Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Tool bracing for performance improvement in simulated femoral head-neck osteochondroplasty Kooyman, Jeremy James Robert 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_kooyman_jeremy.pdf [ 43.12MB ]
Metadata
JSON: 24-1.0074229.json
JSON-LD: 24-1.0074229-ld.json
RDF/XML (Pretty): 24-1.0074229-rdf.xml
RDF/JSON: 24-1.0074229-rdf.json
Turtle: 24-1.0074229-turtle.txt
N-Triples: 24-1.0074229-rdf-ntriples.txt
Original Record: 24-1.0074229-source.json
Full Text
24-1.0074229-fulltext.txt
Citation
24-1.0074229.ris

Full Text

TOOL BRACING FOR PERFORMANCEIMPROVEMENT IN SIMULATED FEMORALHEAD-NECK OSTEOCHONDROPLASTYbyJeremy James Robert KooymanB.Sc, University of Calgary, 2011A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Biomedical Engineering)The University Of British Columbia(Vancouver)August 2013c? Jeremy James Robert Kooyman, 2013AbstractBracing is defined as a parallel mechanical link between a tool user, the environ-ment, and/or the workpiece that alters the mechanical impedance between the tooland workpiece with the goal of improving task performance. Bracing is used in avariety of settings including robotics/automation and more recently in medicine/-dentistry, however it remains relatively understudied in formal ways. This thesisexplored whether bracing could be beneficial in a current orthopaedic problem. Weselected a candidate orthopaedic procedure based on selection criteria that includedthree degrees of freedom, and the ability to abstract/simulate the surgical task usingphantom tests.Femoral head-neck osteochondroplasty is used to treat a deformity of the an-terosuperior femoral head-neck region called cam-type femoroacetabular impinge-ment. During this procedure a surgeon uses a spherical burr to remove the camlesion and restore the normal contour of the femoral head-neck. The goal of thisthesis was to evaluate whether a proposed bracing technique could enable a userto perform a cam resection more accurately and quickly than a currently employedarthroscopic technique.We first performed a pilot study with 4 subjects to examine the impact of brac-ing on simulated bone milling and found that bracing could reduce errors on theorder of 7-14% and procedure length on the order of 30-50% but these findingswere limited by a small sample and effect size. Workspace issues with the braceindicated the need for a redesign, which we combined with the creation of a higherfidelity surgical simulation. We showed that the most effective brace design pro-jected a remote center of motion combined with a spring for axial stiffness.This improved brace design was tested using 20 non-surgeons and 5 surgeons.iiWhile bracing had no detectable effect on the surgeon population, bracing reducedprocedure length and error by 37% and 27% respectively in the non-surgeon pop-ulation when compared to the unbraced condition. Unfortunately, when comparedto the surgical simulation condition, there was no detectable effect of bracing. Thisfinding suggests that an optimal level of bracing may exist but how to experimen-tally determine this level remains a topic for future study.iiiPrefaceThe work presented in this thesis was performed under the supervision of Dr.Antony Hodgson. Dr. Michael Gilbart provided guidance to the work documentedin Chapters 2, 3, and 4.Part of Chapter 2 has been presented at the 13th meeting of the International So-ciety for Computer Assisted Orthopaedic Surgery (computer assisted orthopaedicsurgery (CAOS))Kooyman and Hodgson [2013b] and published in condensed ab-stract form in The Bone & Joint JournalKooyman and Hodgson [2013a], formerlyknown as JBJS(Br). Content in Chapter 3 and all of Chapter 4 has been submittedfor publication.The author, assisted by summer student Yen Po Liu, built the experimentalapparatus, conducted the studies, and performed the data analyses.The human-subject study documented in Chapter 4 were approved by the Uni-versity of British Columbia Behavioural Research Ethics Board (H12-03432) andVancouver Coastal Health Authority (V12-03432).ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xixAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Background on Bracing . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Robotic Bracing Research . . . . . . . . . . . . . . . . . 31.1.2 Human Bracing Research . . . . . . . . . . . . . . . . . 91.1.3 Medicine-Related Bracing Research . . . . . . . . . . . . 91.2 Bracing Potential in Orthopaedics . . . . . . . . . . . . . . . . . 111.3 Femoroacetabular Impingement as a Candidate Application . . . . 131.3.1 The Femoroacetabular (Hip) Joint - Anatomy . . . . . . . 131.3.2 Osteoarthritis of the Hip . . . . . . . . . . . . . . . . . . 141.3.3 Femoroacetabular Impingement . . . . . . . . . . . . . . 15v1.3.4 Femoroacetabular Impingement Epidemiology . . . . . . 161.3.5 Cam-type Femoroacetabular Impingement . . . . . . . . . 171.3.6 Treatment of Cam-type Femoroacetabular Impingement . 191.3.7 Computer-Assisted Femoroacetabular Impingement Inter-ventions . . . . . . . . . . . . . . . . . . . . . . . . . . . 231.3.8 Motivations for Displaying a Clear Boundary for Bone Re-moval . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251.4 Summary and Research Goals . . . . . . . . . . . . . . . . . . . 261.4.1 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . 261.5 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . 262 Preliminary Exploration of Bracing . . . . . . . . . . . . . . . . . . 282.1 Applying Bracing to a Current Orthopaedic Problem . . . . . . . 282.2 Research Objective . . . . . . . . . . . . . . . . . . . . . . . . . 292.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.3.1 Proximal Femur Casting . . . . . . . . . . . . . . . . . . 302.3.2 Laser Scanning . . . . . . . . . . . . . . . . . . . . . . . 302.3.3 Cam Lesion Casting . . . . . . . . . . . . . . . . . . . . 312.3.4 Mock Surgery . . . . . . . . . . . . . . . . . . . . . . . . 322.3.5 Rapidform XOR2/XOV2 Analysis . . . . . . . . . . . . . 342.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.5.1 Study Limitations . . . . . . . . . . . . . . . . . . . . . . 412.6 Conclusions/Future Work . . . . . . . . . . . . . . . . . . . . . . 433 Bracing Apparatus Design . . . . . . . . . . . . . . . . . . . . . . . 443.1 Testing Apparatus Redesign . . . . . . . . . . . . . . . . . . . . 443.1.1 Arthroscope Adaptation . . . . . . . . . . . . . . . . . . 453.1.2 Required Fidelity for Simulation . . . . . . . . . . . . . . 453.1.3 Soft Tissue Interface Design and Integration . . . . . . . . 473.1.4 Testing Sample Redesign . . . . . . . . . . . . . . . . . . 493.2 Exploration of Bracing Design Space . . . . . . . . . . . . . . . . 503.2.1 Fulcrum Brace . . . . . . . . . . . . . . . . . . . . . . . 50vi3.2.2 Armrest Brace . . . . . . . . . . . . . . . . . . . . . . . 513.2.3 Remote Centre of Motion Brace . . . . . . . . . . . . . . 523.2.4 Remote Centre of Motion Brace with Axial Stiffness . . . 533.3 Brace Design Testing . . . . . . . . . . . . . . . . . . . . . . . . 533.3.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 533.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.3.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 624 The Design and Testing of a Bracing Strategy for Simulated FemoralHead-Neck Osteochondroplasty . . . . . . . . . . . . . . . . . . . . 674.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . 704.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 764.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . 785 Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . 795.1 Summary of Thesis Contributions . . . . . . . . . . . . . . . . . 795.2 Applying Bracing to Other Orthopaedic Procedures . . . . . . . . 835.2.1 Arthroscopic Acromioplasty (Shoulder) . . . . . . . . . . 835.2.2 Unicompartmental Knee Arthroplasty . . . . . . . . . . . 845.2.3 The MAKOPlasty versus the Precision Freehand Sculptor 865.3 Closing Thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . 90Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91A VIVID 9i Laser Scanner Specifications . . . . . . . . . . . . . . . . 105B Repeatability Analyses . . . . . . . . . . . . . . . . . . . . . . . . . 106B.1 Scan-Rescan Validation for Proximal Femur Segments . . . . . . 106B.2 Scale Repeatability . . . . . . . . . . . . . . . . . . . . . . . . . 108B.3 Scan-Rescan Validation for Wood Samples . . . . . . . . . . . . . 108viiC Ham Hock Materials Testing System . . . . . . . . . . . . . . . . . . 110C.1 Characterizing Stiffness of Muscle Tissue . . . . . . . . . . . . . 110C.2 Foam Proxy of Muscle Tissue . . . . . . . . . . . . . . . . . . . 111D RCM + Spring Brace Stiffness Optimization . . . . . . . . . . . . . 119D.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119D.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119D.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120D.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120E Light Microscopy Analysis of Spherical Burr Wear . . . . . . . . . 121F Debrief Questionnaire and Responses . . . . . . . . . . . . . . . . . 124F.1 Debrief Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . 124viiiList of TablesTable 2.1 Combined results from the pilot study. . . . . . . . . . . . . . 39Table 3.1 Results from the Mann-Whitney U Test, displayed as p-valuesfrom each pairwise comparison. * indicates significance fol-lowing the Bonferroni-Holm correction for multiple comparisons. 64Table A.1 VIVID 9i Laser Scanner Specifications. c?2006-2013 KonicaMinolta Sensing Americas, Inc. . . . . . . . . . . . . . . . . . 105Table B.1 Raw data from the 45 mesh deviation comparisons made usingRapidform XOV2. . . . . . . . . . . . . . . . . . . . . . . . . 107Table B.2 Data from 30 repetitions of a single weight on a Denver Instru-ment Company AL-500 scale. . . . . . . . . . . . . . . . . . . 108Table B.3 RMS Errors from the 20 reference planes fit to a single samplefor repeatability analysis. . . . . . . . . . . . . . . . . . . . . 109Table C.1 A comparison of the pork and 2040 foam stiffness under vary-ing loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Table C.2 A comparison of the pork and 2050 foam stiffness under vary-ing loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Table C.3 A comparison of the pork and 2050 foam stiffness under vary-ing loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114ixList of FiguresFigure 1.1 During precision tool use, humans will often brace against sur-faces to improve task performance. When writing, rather thanrequiring the arm to co-contract for stability (A), humans willbrace the palm of their hand against the surface (B), makingthe task faster, easier, and more precise. . . . . . . . . . . . . 2Figure 1.2 Bracing is defined as a parallel mechanical link between the ac-tor, the environment, and/or the workpiece that alters the me-chanical impedance between the tool and workpiece in orderto improve task performance. Inset photo illustrating a woodlathe from Jordanhill School D&T Dept. via Wikimedia Com-mons. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Figure 1.3 Some bracing methods for a coarse-fine manipulator. (a) coarsemanipulator (CM) mechanically braced (fine manipulator (FM)free to move), (b) FM virtually braced by task-specific feed-back (CM free to move), (c) FM virtually braced by environ-ment feedback (CM free to move). Hollis and Hammer [1992]c?1992 IEEE . . . . . . . . . . . . . . . . . . . . . . . . . . 5Figure 1.4 Experimental Apparatus of Robot Arm, Large and Flexible(RALF) and Small Arcticulated Arm (SAM) for Hybrid Con-trol. RALF would establish biased contact with a surface, con-trolled by a force sensor in the tip, enabling SAM to performprecision tasks Lew and Book [1994]. c?1994 IEEE . . . . . 6xFigure 1.5 Four types of measurements (a) free robot,(b) human opera-tor,(c) braced robot and (d) human operator with braced fore-arm.Zupanc?ic? and Bajd [1998] . . . . . . . . . . . . . . . . . 7Figure 1.6 Schimmels [2001a] designed a three-point contact jig that as-sisted with edge tracking during deburring operations. By keep-ing all three jig elements in contact with the workpart surfacethrough designed compliance in the manipulator, the debur-ring tool can be controlled via a workpart-edge based referenceframe provided by the jig. . . . . . . . . . . . . . . . . . . . 8Figure 1.7 Laboratory configuration used to explore the objective func-tion of bracing from an ergonomics perspective.Jones et al.[2013] c?2013 Taylor & Francis . . . . . . . . . . . . . . . . 10Figure 1.8 Active Handrest prototype.Fehlberg et al. [2010] c?2010 IEEE 10Figure 1.9 The Johns Hopkins University Steady-Hand robot for cooper-ative human-machine microsurgical manipulation.Taylor et al.[1999] c?1999 Sage Publications Ltd. . . . . . . . . . . . . . 12Figure 1.10 Posterior view of right proximal femur. Grey [1918] . . . . . 14Figure 1.11 Lateral view of right hemi-pelvis. Grey [1918] . . . . . . . . 15Figure 1.12 Anterior view of right femoroacetabular joint with intact jointcapsule. Grey [1918] . . . . . . . . . . . . . . . . . . . . . . 16Figure 1.13 Lateral view of right femoroacetabular joint with view of theanterosuperior femoral head-neck region. Grey [1918] . . . . 17Figure 1.14 Cam lesion indicated by bony deformity at the anterolateralhead-neck junction (A), which impinges on the labrum andarticular cartilage of the acetabulum (B). Reprinted with per-mission, J.W.Thomas Byrd, MD, Nashville Sports MedicineFoundation, www.nsmfoundation.org. Byrd and Jones [2009] 18Figure 1.15 anterior posterior (AP) radiograph showing a patient with cam-type impingement. Bredella et al. [2013] c?Elsevier 2013 . . 19Figure 1.16 Operating room setup for supine arthroscopic treatment ofFemoroacetabular impingement (FAI). Dienst [2005] . . . . . 21Figure 1.17 Placement of portals for the supine arthroscopic approach. Di-enst [2005] . . . . . . . . . . . . . . . . . . . . . . . . . . . 22xiFigure 1.18 Arthroscopic view of a cam lesion (A). Spherical burr perform-ing osteoplasty (B). FH is femoral head. OST is osteoplasty.Ejnisman [2011] . . . . . . . . . . . . . . . . . . . . . . . . 23Figure 1.19 A patient with a normal alpha angle less than 50? (A) and apatient with a cam lesion at the femoral head-neck region, withalpha angle measuring 65? (B). Dimmick et al. [2013] . . . . 24Figure 2.1 A visual overview of the stages in the pilot study. . . . . . . . 29Figure 2.2 A silicone mold (A) was created from the proximal segment ofa commercially available adult left femur. 16 Proximal femursegments were cast in this mold (B) using white urethane plastic 30Figure 2.3 A silicone mold showing a cam lesion created by pouring blackurethane plastic in from a hole in the exterior of the mold, pro-viding a perfect fit to the surface of the proximal femur segment. 32Figure 2.4 Cam lesions cast from black urethane plastic. It can be seenthat a learning curve is present in the two-stage casting method,however only appropriate samples such as the far right wereused in experimentation. . . . . . . . . . . . . . . . . . . . . 33Figure 2.5 The mock surgical setup for simulated femoral head-neck os-teochondroplasty. Subjects would use a handheld arthroscopyshaver with 5.5mm spherical burr to remove the black cam le-sion with or without the aid of the spherical bearing tool brace,placed in the approximate position of the anterolateral portal. . 34Figure 2.6 Mesh-rebuild process in Rapidform XOR2. . . . . . . . . . . 35Figure 2.7 3d models of the proximal femur segments were trimmed tothe region of interest (ROI) to prevent influence of extraneousinformation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 36Figure 2.8 The RMS error in Rapidform XOV2 is determined by project-ing the post-resection data onto the pre-resection data, thencalculated the difference between the two surfaces via a pro-jection to the closest point. . . . . . . . . . . . . . . . . . . . 36xiiFigure 2.9 A visual representation of the RMS error in Rapidform XOV2.Thinking of the colours as a depth map, the positive effects ofbracing are evident. . . . . . . . . . . . . . . . . . . . . . . . 37Figure 2.10 Results from the simulated femoral head-neck osteochondroplasty,with subjects being told to be as fast as possible with moderaterespect for accuracy. RMS error (mm) is on the x-axis, andprocedure length (s) is on the y-axis. Hollow shapes representthe unbraced condition, and filled shapes represent the bracedcondition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Figure 2.11 Results from the simulated femoral head-neck osteochondroplasty,with subjects being told to be as accurate as possible with mod-erate respect for time. RMS error (mm) is on the x-axis, andprocedure length (s) is on the y-axis. Hollow shapes representthe unbraced condition, and filled shapes represent the bracedcondition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Figure 2.12 Combined results from the simulated femoral head-neck os-teochondroplasty. RMS error (mm) is on the x-axis, and pro-cedure length (s) is on the y-axis. Hollow shapes representthe unbraced condition, and filled shapes represent the bracedcondition. Data is presented as the mean +/- standard deviation. 40Figure 2.13 A hypothetical categorization of existing orthopaedic surgicaltechnology on a time-error plane with the ideal scenario of zerotime and zero error interventions displayed in the bottom leftcorner. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Figure 2.14 Three of the four subjects used in the study, each displayingdifferent task postures. (A) is bracing against the workpiece,(B) is bracing against their trunk, and (C) is bracing against thetarget by shortening the effective tool tip length. . . . . . . . . 42Figure 2.15 The spherical bearing tool brace did not provide sufficient workspaceto allow for the entire cam lesion to be removed. . . . . . . . 42xiiiFigure 3.1 Arthroscopes display their field of view at an angle to the axisof the scope, denoted by ? . Typical arthroscopes used in FAIinterventions are 30? and 70?, although the latter is used pre-dominantly during acetabular procedures. . . . . . . . . . . . 45Figure 3.2 A comparison of a 30? arthroscope (A, C) against our proposedUSB adaptation (B, D), showing an unresected sample (A, B)and a resected sample (C,D). . . . . . . . . . . . . . . . . . . 46Figure 3.3 A hip/femur model was inserted into a mannequin and coveredwith a foam surface to allow for palpation of bony landmarksused to triangulate arthroscopic portal locations. . . . . . . . . 48Figure 3.4 An overview of the soft tissue simulation used in the presentexperiment. 2050 foam and 5mm thick closed-cell foam (A)were placed in a wood box, forming the soft tissue support(STS) condition, which housed the 3D printed sample holder(B). The capsulotomy, which joined the tool and camera por-tals, is shown as a slit in (B). The distance from the capsule tothe sample, shown in (C) was based off of the approximate in-sertion depth of the tool as observed during surgery. The finalbox with soft tissue interface, is shown in (D) with the fulcrumbrace in place to highlight the anterior portal. . . . . . . . . . 49Figure 3.5 Mock cam lesion used in the present study. Black plastic shapeglued onto wood substrate. . . . . . . . . . . . . . . . . . . . 50Figure 3.6 The fulcrum brace, consisting of a spherical bearing rod-end,was mounted at the approximate location of the tool portal forthe present study. . . . . . . . . . . . . . . . . . . . . . . . . 51Figure 3.7 The armrest, which kept the subject?s elbow at approximately90? flexion, used in the present study. . . . . . . . . . . . . . 52Figure 3.8 The remote centre of motion (RCM) was placed in a locationthat projected the instantaneous centre of rotation along thetool insertion vector required by the surgical approach and thefulcrum brace, at a depth of 2X the insertion depth. . . . . . . 54Figure 3.9 The RCM brace from Figure 3.8 shown here with a spring thatadded axial stiffness to the tool. . . . . . . . . . . . . . . . . 55xivFigure 3.10 (A) A closeup of the RCM brace from Section 3.2.3, and (B)the RCM + Spring brace from Section 3.2.4, highlighting thedifference between the two designs. . . . . . . . . . . . . . . 55Figure 3.11 Small amounts of black lesion, under 1 mm2 as visually as-sessed by the facilitator, can remain after the mock surgeryprovided the subject feels that they have adequately resectedthe cam lesion. In the top two cases, the subject would be al-lowed to continue the resection. In the bottom two cases, theresection would be considered complete. . . . . . . . . . . . . 57Figure 3.12 A custom three dimensional (3D) printed sample holder wasused to keep each sample in a repeatable position during laserscanning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Figure 3.13 Laser scans are trimmed of extraneous data (A) and aligned(B-C) and merged into a final mesh (D). . . . . . . . . . . . . 59Figure 3.14 The edges of each sample (A) are trimmed (B) in RapidformXOR2 to reduce extraneous data and improve the fit of thereference plane and resulting mesh deviation calculation. . . . 59Figure 3.15 A reference plane (A) is fit to a sample by selecting severalpoints on the surface of the sample unaffected by the cam le-sion resection. A side view is shown in (B). . . . . . . . . . . 60Figure 3.16 The RMS error in Rapidform XOV3, using the reference planemethod, is determined by projecting the post-resection dataonto the reference plane data, then performing the calculationbased on the mesh deviation error. . . . . . . . . . . . . . . . 61Figure 3.17 Two examples of the mesh deviation visualization from Rapid-form XOV3. The deeper the blue, the greater the amount ofover resection that has occurred. . . . . . . . . . . . . . . . . 62Figure 3.18 Combined results showing the performance of the differentbrace designs compared to an unbraced control. . . . . . . . . 63Figure 3.19 A comparison of typical scope views during the STS (left) andunbraced conditions (right). . . . . . . . . . . . . . . . . . . 64xvFigure 3.20 A conceptual comparison of the brace designs presented so farin this study, mapped on a chart comparing motion type on thex-axis and effective stiffness at the hand on the y-axis. . . . . 65Figure 4.1 Two styles of handwriting. Unsupported (A) and supportedthrough palm contact on the writing surface (B). . . . . . . . 69Figure 4.2 The simulated cam lesion used in the study. The poplar woodsubstrate is 1.5x1.5. The black representative cam lesion is19mm in diameter, 3.6mm thick at the centre, with a weight of0.6 +/- 0.05 g. . . . . . . . . . . . . . . . . . . . . . . . . . . 70Figure 4.3 Subjects were tested under three different conditions: (A) FreeCutting - The tool received no support. (B) Soft Tissue Support(STS) - The test box was filled with a low and high densityfoam to simulate the support provided by soft tissue and thejoint capsule. (C) Braced The STS condition was combinedwith a mechanical brace that eliminated the reversal of motionof the tool that occurs in the STS condition, as well as providedaxial stiffness through a compression spring, making the toolstiffer as it approached the sample. . . . . . . . . . . . . . . . 71Figure 4.4 Rapidform XOR2 was used to process raw scan data into a pla-nar 3D model. Scan data was loaded using the mesh buildupfeature (A), aligned using a globally referenced iterative clos-est point algorithm (B), merged into a single model (C), andfinalized by redistributing the poly faces (D). The final modelused for comparison is shown with manually trimmed edgesin (E). Rapidform XOV3 was used to analyze the RMS rough-ness as calculated by a plane fit to the surface of the sample(F). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Figure 4.5 Results are shown as the mean +/- SD, compared to the idealsurgical intervention in the bottom left corner. . . . . . . . . . 74Figure 4.6 Results comparing the average deviation as a function of thesupport condition, shown as the mean +/- SD, compared to theideal surgical intervention at the dashed horizontal axis. . . . 75xviFigure 5.1 The left shoulder and acromioclavicular joint. The muscula-ture is not shown. Grey [1918] . . . . . . . . . . . . . . . . . 84Figure 5.2 The left shoulder with musculature, emphasizing the rotatorcuff and the supraspinatus muscle superior to the spine of thescapula. Grey [1918] . . . . . . . . . . . . . . . . . . . . . . 85Figure 5.3 Sagittal view of the shoulder, with arthroscopic milling tool en-tering posteriorly to remove the shaded portion of the acromion.Sampson et al. [1991] . . . . . . . . . . . . . . . . . . . . . . 86Figure 5.4 The MAKO robotic arm interactive orthopaedic system (RIO)(A) and the NavioPFS (B) being used by the thesis author atthe 2013 International Meeting of the International Society forComputer Assisted Orthopaedic Surgery. . . . . . . . . . . . 87Figure B.1 Whole-deviation map from Rapidform XOV2 . . . . . . . . . 107Figure C.1 A 22lb pork hock purchased from a local butcher. The skin-fat-fascia-muscle interface is clearly visible from the side. . . 112Figure C.2 The material testing system used to quantify the stiffness ofporcine tissue and foam proxies. It uses a system of pulleys toconvert a downwards force into a horizontal one to apply to thesample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Figure C.3 Results from testing the stiffness of pork tissue. Time in sec-onds is on the x-axis, with displacement in meters on the y-axis. According to Hooke?s law, under 5lbs the tissue stiffnesswas 3175.8 N/m, under 10lbs the tissue stiffness was 4170.6N/m, and under 15lbs the tissue stiffness was 5647.2 N/m . . 114Figure C.4 Results from testing the stiffness of 2040 foam. Time in sec-onds is on the x-axis, with displacement in meters on the y-axis. According to Hooke?s law, under 5lbs the foam stiffnesswas 1953.9 N/m, under 10lbs the foam stiffness was 2495.3N/m, and under 15lbs the foam stiffness was 2056.4 N/m . . . 115xviiFigure C.5 Results from testing the stiffness of 2050 foam. Time in sec-onds is on the x-axis, with displacement in meters on the y-axis. According to Hooke?s law, under 5lbs the foam stiffnesswas 2482.6 N/m, under 10lbs the foam stiffness was 3045.6N/m, and under 15lbs the foam stiffness was 3399.2 N/m . . . 116Figure C.6 Results from testing the stiffness of blue foam. Time in sec-onds is on the x-axis, with displacement in meters on the y-axis. According to Hooke?s law, under 5lbs the foam stiffnesswas 3249.1 N/m, under 10lbs the foam stiffness was 5265.6N/m, and under 15lbs the foam stiffness was 5223.4 N/m . . . 117Figure C.7 A visual comparison of the various foam properties when cor-rected for thickness, illustrating that the 2050 foam providedproperties closest to pork tissue. . . . . . . . . . . . . . . . . 118Figure D.1 Results comparing the impact of spring stiffness on time andaccuracy of lesion removal. . . . . . . . . . . . . . . . . . . . 120Figure E.1 Light microscopy was used to compare the cutting surfaces ofa brand new (top) and used (bottom) 5.5mm abrader burr. . . . 123Figure F.1 Pooled results (Surgeons and Non-Surgeons) from the debriefquestionnaire. Results were pooled due to the small Surgeongroup (n=5) and the same trends being identified in each group. 127xviiiGlossary2D two dimensional3D three dimensionalAP anterior posteriorASIS anterior superior iliac spineCAD computer-aided designCAS computer assisted surgeryCAOS computer assisted orthopaedic surgeryCM coarse manipulatorCT computer tomographyDOF degree of freedomEMG electromyographyFAI Femoroacetabular impingementFEM finite element methodFM fine manipulatorGUI graphical user interfaceMTS material testing systemxixOA osteoarthritisRALF Robot Arm, Large and FlexibleRCM remote centre of motionRIO robotic arm interactive orthopaedic systemRMS root mean squareROI region of interestROM range of motionSAM Small Arcticulated ArmSTS soft tissue supportTHA total hip arthroplastyTKA total knee arthroplastyUBC University of British ColumbiaUKA unicompartmental knee arthroplastyxxAcknowledgmentsI would like to thank my supervisor, Dr. Tony Hodgson, who granted me thefreedom to explore this research area. It has been a rewarding experience and hisguidance was key to my success.Many students with the Neuromotor Control Laboratory have provided me withsupport and feedback throughout these past two years. I would like to express mygratitude to Jake McIvor, Tarek Mohammad, and Mark Semple for taking time outof their work to assist with mine. Yen Po Liu, the summer student that assisted mein 2013, is worthy of a special mention as I could not have made it without him.The staff at the Centre for Hip Health and Mobility consistently went aboveand beyond to assist with and solve technical problems. I would be in a muchdifferent place without their support.Further thanks is extended to those who volunteered their time to participate inmy research studies, and the Natural Sciences and Engineering Research Councilof Canada for their financial support.xxiDedicationTo my family,To the parents who never stopped believing in me.To the sisters who drove me to push harder and reach farther.From strong roots rise mighty trees.xxiiChapter 1IntroductionThis chapter provides an overview of the background and motivations for thisstudy, giving context to the project objectives. We then give an overview of theremaining thesis chapters.1.1 Background on BracingBracing is a strategy that people use either intentionally or subconsciously that canimprove performance in a variety of tool-using tasks. This bracing strategy is anal-ogous to people who brace their hand against a surface during writing, as shownin Figure 1.1, with subsequent precision motion provided by only the degrees offreedom in the hand, rather than the entire arm. Bracing can be more generallydescribed as a parallel mechanical link between the actor, the environment, and/orthe workpiece that alters the mechanical impedance between the tool and work-piece in order to improve task performance, as illustrated in Figure 1.2. Such alink creates a secondary, parallel load path which can increase end-effector stabil-ity, stiffness, and may reduce fatigue if a person is performing the task instead of arobot manipulator.Literature on bracing is limited, and focuses on three main areas:1. Applying bracing to improve the performance of robotic arms2. Effects on ergonomics of human-tool interaction1Figure 1.1: During precision tool use, humans will often brace against sur-faces to improve task performance. When writing, rather than requiringthe arm to co-contract for stability (A), humans will brace the palm oftheir hand against the surface (B), making the task faster, easier, andmore precise.Figure 1.2: Bracing is defined as a parallel mechanical link between the ac-tor, the environment, and/or the workpiece that alters the mechanicalimpedance between the tool and workpiece in order to improve task per-formance. Inset photo illustrating a wood lathe from Jordanhill SchoolD&T Dept. via Wikimedia Commons.23. Effects on performance in clinical applications1.1.1 Robotic Bracing ResearchOne of the first published articles on bracing was by Book et al. [1984], whichpresented a strategy to improve the performance of lightweight robotic arms. Theyproposed that an arm could be ?moved into position, then rigidized by bracingagainst either the workpiece or an auxiliary, static structure.? Methods consid-ered for bracing a robot arm included preloading against a surface, mechanical andmagnetic connections, and vacuum clamping. A later publication, Book [1984],reiterated the concepts presented by Book et al. [1984] and bring to light some ofthe difficulties in controlling braced robot arms during gross and fine motion, andrendezvous/inactive phases. These were the first publications to propose bracingas a method for improving precision of robotic end effectors.Asada and West [1985] and Asada and Sawada [1985] similarly presented brac-ing as a method for improving stiffness of robotic end effectors, in the context of anadaptable tool guide for automated grinding robots. Weld-seam grinding requiredmachining tolerances of 0.1-0.5mm but conventional approaches were subject to anumber of uncertainties and disturbing influences including:1. Reaction forces generated between the grinding tool and workpiece2. Reaction forces varying as a function of the end point velocity of the robotarm3. Workpiece tolerancing preventing the precise location of the work surfacefrom being known to the robotThe application of the adaptable tool guide to the end of an industrial grinding robotincreased the stiffness of the end effector by a factor of 50 in the normal directionof the work surface over the conventional configuration. It was not mentioned ifthe adjustable tool guide impacted overall performance.West [1986] provided guidelines to design mechanisms for braced manipula-tors. In this case, ?design? is used to encompass the decision to adopt a bracingstrategy, the design of the brace/jig hand and the selection of the control method-3ology. West recommended adopting a bracing strategy when manipulator perfor-mance was unsatisfactory because of:1. Insufficient load capacity2. Insufficient stiffness3. Insufficient position resolution4. Excessive manipulator costAdditionally, a task is available to bracing when:1. There are suitable bracing surfaces available, especially if the load is directedtowards the bracing surface, and/or2. The task can be easily segmented into coarse and fine manipulation and pre-cision tasksThese characteristics make it more likely that a bracing strategy will increase ma-nipulator performance and/or reduce its cost.Kwon and Book [1988] examined the coarse-fine manipulation concept pre-sented by West [1986] calling the idea ?staged positioning?, a generalized termwith a braced manipulator as one example. They examined the effects of bracingfrom the view of uncertainty or error reduction, and they demonstrated that bracingappeared to be effective at reducing random (sensor resolution, gear backlash, andbearing looseness) and vibration (flexible beam and joint) errors.A parametric comparison of hypothetical unbraced and braced manipulatorswas performed by Book and Wang [1989], who examined gross and fine mo-tion ranges during a pick-and-place task. Braced arms were shown to be supe-rior to unbraced under some combinations of task parameters, particularly in largeworkspaces where the coarse manipulator can cover a large distance quickly andfor large numbers of part moves (n=50), highlighting the advantage of bracing forfine manipulation tasks.The concept of virtual bracing (i.e. bracing provided by a control system ratherthan mechanical means) was studied by Hollis and Hammer [1992]. They built on4Figure 1.3: Some bracing methods for a coarse-fine manipulator. (a) CM me-chanically braced (FM free to move), (b) FM virtually braced by task-specific feedback (CM free to move), (c) FM virtually braced by environ-ment feedback (CM free to move). Hollis and Hammer [1992] c?1992IEEE5Figure 1.4: Experimental Apparatus of RALF and SAM for Hybrid Control.RALF would establish biased contact with a surface, controlled by aforce sensor in the tip, enabling SAM to perform precision tasks Lewand Book [1994]. c?1994 IEEEprevious work by Kwon and Book [1988] West [1986] examining physical brac-ing for coarse manipulation (CM) and virtual bracing for fine manipulation (FM),with examples shown in 1.3. Using precision assembly of stacked laminates by acoarse-fine manipulator, bracing the CM provided an order of magnitude assem-bly error reduction when compared to an unbraced coarse-fine manipulator, andalmost two orders of magnitude better than coarse manipulation alone, demon-strating that bracing results in improved position accuracy and stiffness. In anuninterrupted 26 hour laminate assembly run, the braced assembly had 95% ofplacements within 2.5 ?m compared to a traditional assembly system with 25 ?m-50 ?m accuracy. They were the first group to propose the idea of physically bracingthe CM while using virtual references to ?brace? the FM.Lew and Book [1994] presented a bracing strategy for micro/macro manipu-lators, akin to the CM/FM bracing from Hollis and Hammer [1992], however withmore focus on the controller design. They used a manipulator with two stages,RALF (Robotic Arm, Large and Flexible), and SAM (Small Articulated Arm) witha bracing foot as shown in 1.4, to show that bracing causes a change in structuralvibrations in flexible manipulators. Specifically, bracing RALF produced higherfrequency and smaller amplitude structural vibrations, suggesting increased end-6Figure 1.5: Four types of measurements (a) free robot,(b) human operator,(c)braced robot and (d) human operator with braced forearm.Zupanc?ic? andBajd [1998]effector precision.Zupanc?ic? and Bajd [1998] tested the position repeatability of robot and humanmanipulators with and without a bracing strategy in place, as shown in Figure1.51.Using the ISO 9238 repeatability measurement standard, applied for the first timeto a human operator, they showed a 50% improvement in robot and 25% improve-ment in human position repeatability, suggesting that further exploration shouldoccur in ergonomics and robotics fields for improvement in accuracy, repeatability,stiffness, payload capacity and vibration compensation.Schimmels [2001a] and Schimmels [2001b] developed a method for improv-ing end-effector positioning capability and increasing effective stiffness using abracing strategy.1Reprinted from Computers in biology and medicine, 28/4, Zupancic and Bajd, Comparison ofposition repeatability of a human operator and an industrial manipulating robot, 415-421, Copyright(1998), with permission from Elsevier7Figure 1.6: Schimmels [2001a] designed a three-point contact jig that as-sisted with edge tracking during deburring operations. By keeping allthree jig elements in contact with the workpart surface through designedcompliance in the manipulator, the deburring tool can be controlled viaa workpart-edge based reference frame provided by the jig.Figure 1.62 illustrates an end-effector mounted jig capable of establishing aworkpart-based reference frame to conduct edge deburring operations. The physi-cal constraints of the task environment are used to provide relative positional infor-mation of the manipulator end-effector with respect to the workpart, hypotheticallyimproving robot performance.SummaryBracing, as applied to robotic manipulators, has the capacity to improve end-effector stiffness, load capacity, and position resolution, compensate for structuralvibrations and tool-workpiece interaction forces, and provide relative end-effectorpositional information to the controlled. Separately or combined, these featurescan improve task performance without requiring a redesign of the manipulator.Whether potential performance improvements justify the design of a brace is likelytask-dependent, with the availability of bracing surfaces playing a large role indesign complexity.2Reprinted from Robotics and Computer-Integrated Manufacturing, 17/4, Schimmels, Joseph,Multidirectional compliance and constraint for improved robotic deburring. Part 1: improved posi-tioning, 277-286, Copyright (2001), with permission from Elsevier81.1.2 Human Bracing ResearchAs suggested in 1.1, humans naturally adopt bracing strategies in a variety of dif-ferent tasks to improve performance. Jones et al. [2008] studied the behaviourin an automotive assembly plant setting and hypothesized that subjects adoptedbracing postures to reduce and/or change the tasks demands. Of all the tasks theyexamined, 48% were performed with some degree of bracing behaviour, rangingfrom the contralateral hand to establishing thigh contact. They suggest that theobjective function of bracing is to enable effective recruitment of body weight tocomplete a given force exertion, and propose that both external bracing (body con-tact) and contralateral one-hand support increase force exertion capabilities. Laterwork would analyse one-hand maximal push, pull and lift tasks and demonstratethat bracing surfaces available to the thighs and non-task hand enabled participantsto exert an average of 43% more force at the task hand (Jones et al. [2010] Joneset al. [2013]).Fehlberg et al. [2010] and Fehlberg et al. [2012] explored bracing of the coarsemanipulator provided by a human arm to address the limited dextrous workspaceof the hand through the use of an active handrest shown in Figure 1.8. Participantswere asked to draw circles using an Omni stylus that rendered a virtual writingsurface linked to a graphical user interface (GUI), balancing speed and accuracy.The Active Handrest lead to a 26% reduction in error compared to a fixed handrestbut with significantly longer completion times. This suggests that there may be amore optimal bracing solution for the given task.1.1.3 Medicine-Related Bracing ResearchDentistryA tour of the University of British Columbia (UBC) Dentistry facilities, [L. Rucker(personal communication, March 12, 2012)], demonstrated that trainees are taught,from an ergonomics perspective, to establish hand rests on a patient?s face and fin-ger rests inside a patient?s mouth to aid with tool precision and to reduce the ***po-tential for the development of cumulative trauma disorders in dental professionals.Research from Dong et al. [2005] examined hand muscle activity during dental9Figure 1.7: Laboratory configuration used to explore the objective functionof bracing from an ergonomics perspective.Jones et al. [2013] c?2013Taylor & FrancisFigure 1.8: Active Handrest prototype.Fehlberg et al. [2010] c?2010 IEEE10scaling and reported that the use of finger rests reduced hand muscle activity andpinch force, excessive amounts of which have been suggested to lead to cumula-tive trauma disorders. While the authors report that finger stabilization throughestablishing finger rests improves the precision of scaling, they did not explicitlyexamine or test this statement. Cosaboom-FitzSimons et al. [2008] performed afollow-up study, noting that the Dong et al. [2005] study had questionable meth-ods, especially the susceptibility of cross-talk with their electromyography (EMG)setup and the limited testing workspace (one quadrant of the mouth on one tooth).They reported that finger rests offer minimal ergonomic advantages in terms ofmuscle activity and types of fulcrums used, contrary to the study by Dong et al.[2005]. Discussions with a dental hygienist, [C. Ness (personal communication,June 14, 2012)], indicate that the practice of finger rests is extensively taught inhygienist education as a method of improving tool precision and ergonomics, de-spite the lack of conclusive studies in the literature.MicrosurgerySurgery on structures found in the eye require micron-scale tolerances and place ahigh demand on a surgeon?s ability. Taylor et al. [1999] designed the Steady-Handrobot to augment microsurgical performance by working cooperatively with thesurgeon. They examined the ability of the human operator to position a microsur-gical needle to 250, 200, and 150 ?m accuracy. The Steady-Hand system improvedsuccess rates from 49% unassisted to 79% for 150 ?m holes and from 49% unas-sisted to 78% unassisted for 250 ?m holes. The use of cooperative control, sensingthe surgeon?s input and scaling the force to provide precise, tremor-free positioncontrol, is one of the first examples of cooperative medical robotics.1.2 Bracing Potential in OrthopaedicsFrom section 1.1, bracing in different forms has the potential to improve sta-bility and precision of tools (Book et al. [1984]Book [1984]Asada and Sawada[1985]Asada and West [1985]West [1986]Book and Wang [1989]Hollis and Ham-mer [1992]Taylor et al. [1999]Schimmels [2001a]Schimmels [2001b]Fehlberg et al.[2010]Fehlberg et al. [2012]) , reduce the effects of perturbations from vibrations11Figure 1.9: The Johns Hopkins University Steady-Hand robot for cooperativehuman-machine microsurgical manipulation.Taylor et al. [1999] c?1999Sage Publications Ltd.and workpiece interactions (Kwon and Book [1988]Lew and Book [1993]) , im-prove position repeatability of an end-effector (Zupanc?ic? and Bajd [1998]Schim-mels [2001a]Schimmels [2001b]) , and enable greater force production than anunbraced configuration (Jones et al. [2008]Jones et al. [2010]Jones et al. [2013]),without drastically increasing cost.Since the demand for procedures is rising (Kurtz et al. [2009] Fehring et al.[2010] Bombardier et al. [2011]Montgomery et al. [2013]), surgeons are naturallyreluctant to embrace technologies that require increased procedure time (eg. mostcomputer assisted orthopaedic surgery (CAOS) technologies). Rivkin and Lieber-gall [2009] reported several barriers to CAOS adoption including ?user-friendliness?,learning curves that affect all members of the operating team, ergonomic factorssuch as the space requirements of the surgical system, and the large capital cost ofCAOS systems.Bracing could potentially provide a low-cost, intuitive method of improvingperformance of robotics and humans in a range of tasks but it is relatively under-studied in medicine applications, particularly the field of orthopaedics. This studysought to explore whether bracing could be beneficial in a current orthopaedicproblem.An ideal candidate procedure would have the following characteristics:121. Three degree of freedom (DOF) for the tool to provide a logical extension ofprevious one DOF work from our lab.2. Suitable surgical complexity to enable abstraction and testing on a phantommodel.3. The ability to be tested using both a surgeon and non-surgeon population.1.3 Femoroacetabular Impingement as a CandidateApplicationFemoroacetabular impingement (FAI) is a source of hip pain in a young popu-lation and has been recently shown to cause early-onset osteoarthritis (OA) in aprospective cohort study (Agricola et al. [2012]) after years of accumulating ev-idence supporting this hypothesis (Ganz et al. [2003] Ganz et al. [2008] Leuniget al. [2009]). Research has described three types of FAI: cam, pincer, and mixed/-combined (a combination of the two former types). Cam impingement is the resultof a non-spherical portion of the femoral head impacting the articular surface ofthe acetabulum, resulting in reduced range of motion (ROM) and failure of the ar-ticular surface of the joint. It has been shown that the accuracy of resection of thecam lesion impacts the surgical outcome (Philippon et al. [2007] Sampson [2001]Willimon et al. [2011]) and that revision hip arthroscopy is caused by inadequateresection of the cam lesion( Heyworth et al. [2007] Papavasiliou and Bardakos[2012]), indicating a need for more accurate resection. To meet a rising demandfor FAI treatment, the objective of this study was to design a bracing strategy thatenabled the used to be better and faster at performing an arthroscopic milling taskthat simulated FAI treatment than currently employed orthopaedic techniques.1.3.1 The Femoroacetabular (Hip) Joint - AnatomyThe hip is a synovial joint formed by the head of the proximal femur (Figure 1.10),articulating with the acetabulum of the pelvis (Figure 1.11), surrounded by a syn-ovial membrane and joint capsule shown in Figure 1.12. Both joint surfaces arecovered with articular cartilage, which serves to support the primary connectionbetween the trunk and lower limbs in a human. The spherical head of the femur13Figure 1.10: Posterior view of right proximal femur. Grey [1918]and the cup-like acetabulum form a ball-and-socket joint designed for stability overa wide range of movement. During standing and locomotion, the entire weight ofthe upper body is transmitted through the hip joint to the heads and necks of thefemurs.1.3.2 Osteoarthritis of the HipOA is a degenerative joint disease with epidemiological studies estimating it im-pacts 1 in 8 Canadians, and 80% of the global population above 75 years of age(Bombardier et al. [2011] Arden and Nevitt [2006]). With aging or acute injury,the smooth, healthy cartilage becomes roughened and eroded leading to pain anddisability.While the incident rate of OA is expected to increase with an ageing popula-tion confounded by the obesity epidemic in North America (Stu?rmer et al. [2000]Fehring et al. [2010]) research has shown that early onset OA can be found in ayoung population afflicted with a pathomechanical process caused by bony im-pingement of the femur and the acetabulum, called FAI (Ganz et al. [2003]). Earlydetection and treatment of the deformity can mitigate degenerative damage to thehip and prevent the onset of the disease (Agricola et al. [2012]).If no changes are made in preventing the incidence rate of OA, by 2020 it isestimated that the Canadian cumulative total economic burden, including direct14Figure 1.11: Lateral view of right hemi-pelvis. Grey [1918]and indirect costs, of the disease will reach $405.1 billion, up from $27.5 billion in2010 (Bombardier et al. [2011]).1.3.3 Femoroacetabular ImpingementThe concept of FAI was first mentioned by Elmslie [1933] who postulated thatpeople who developed OA by the relatively early age of 40-50 may have had anundiagnosed hip joint deformity. Murray [1965] would revisit this concept, hy-pothesizing that most primary hip OA cases were secondary to subtle anatomicalvariations that were within normal radiological limits. He would describe a ?tiltdeformity? of the femoral head in relationship to the femoral neck and after re-viewing 200 anterior posterior (AP) radiographs of patients with OA,he concludedthat the deformity was present in 39.5% of cases.The findings by Murray [1965] would be supported by Stulberg and Harris[1974],Stulberg et al. [1975], and Solomon [1976], so by the mid-1970s the con-cept that mechanical impingement from subtle femoroacetabular deformity was the15Figure 1.12: Anterior view of right femoroacetabular joint with intact jointcapsule. Grey [1918]pathomechanism responsible for many cases of pain in the hip was widely adoptedby researchers (Lavinge et al. [2010]). Advancements in clinical tests, imagingtechnology, and surgical techniques would enable Ganz et al. [2003] to proposethat FAI played a role in the development of OA, based on a study of more than600 cases undergoing surgical hip dislocation. Since this study, there has been anexponential increase in publications on FAI.1.3.4 Femoroacetabular Impingement EpidemiologyFAI has limited epidemiological data (Lung et al. [2012]), however it is estimatedto be present in 10-15% of the population. A retrospective study by Jung et al.[2011] looked at computer tomography (CT) scans of 215 male and 540 femalehips and found that 13.95% of male and 5.56% of female patients were consideredto have FAI despite being asymptomatic.Due to the lack of large clinical FAI studies, it is currently the target of theANCHOR FAI-1 clinical study (Clohisy et al. [2013]) which seeks to investigate16Figure 1.13: Lateral view of right femoroacetabular joint with view of theanterosuperior femoral head-neck region. Grey [1918]the contemporary treatment of FAI. Descriptive reports from the study cohort (1076patients, 1130 hips) show that the mean age of patients was 28.4 years. This isconsistent with findings from Lung et al. [2012] who reported that FAI was commonin young patients undergoing total hip arthroplasty (THA), *** evidenced by 36%of their subjects being younger than 55 years.1.3.5 Cam-type Femoroacetabular ImpingementFAI is characterized by an abnormal morphology of the hip leading to mechani-cal impingement of the proximal femur against the acetabulum during flexion andinternal rotation, with repeated contact leading to the development of symptoms(Ganz et al. [2008]).Two distinct types of FAI have been identified (Ganz et al. [2008]): The jam-ming of a nonspherical osseous growth on the anterosuperior femoral head/neckinto the acetabulum, named cam FAI (Figure:1.14), and an overcoverage of theacetabulum leading to impingement on the head-neck junction, named pincer FAI.17Figure 1.14: Cam lesion indicated by bony deformity at the anterolat-eral head-neck junction (A), which impinges on the labrum andarticular cartilage of the acetabulum (B). Reprinted with permis-sion, J.W.Thomas Byrd, MD, Nashville Sports Medicine Foundation,www.nsmfoundation.org. Byrd and Jones [2009]During cam FAI, the lesion impacts the acetabular cartilage leading to progres-sive degradation and OA as shown in Figure1.14 and discussed in Section 1.3.2.For reference, a normal femoral head-neck is shown in Figure 1.13, and surgeonsattempt to correct to this normal curvature through surgical intervention.18Figure 1.15: AP radiograph showing a patient with cam-type impingement.Bredella et al. [2013] c?Elsevier 20131.3.6 Treatment of Cam-type Femoroacetabular ImpingementThere is both non-operative and operative treatment for FAI. Nonoperative treat-ment, including activity modification, non-steroidal anti-inflammatory medicationsand intra-articular injections, have been shown to have limited success (Zebalaet al. [2007] Kivlan et al. [2011]). Conversely, surgical treatment, specificallyarthroscopy, has been shown to benefit FAI patients, with Cooper et al. [2012]reporting a significant increase in the Harris Hip Score, a scale that covers thepain, function, absence of deformity, and ROM domains (Nilsdotter and Breman-der [2011]).Operative treatment offers several approaches ranging from open surgery (Botseret al. [2011] Papalia et al. [2012]), combined limited open (Zebala et al. [2007]),and arthroscopic surgical approaches ( Ganz et al. [2003] Kendoff et al. [2011]).Of these options, arthroscopy offers reduced post-operative morbidity, and shorterrehabilitation time, at the expense of being technically demanding (Philippon et al.19[2007] Botser et al. [2011] Papalia et al. [2012]).During arthroscopic treatment of cam-type FAI, portals into the hip joint arecreated and long, slender tools are inserted. The joint is visualized with an arthro-scope, and the cam lesion is removed using a hand-held burring tool, with a typicalbit size of 5.5mm. There are two main arthroscopic surgical approaches, lateraland supine, referring to the position of the patient on the operating table.The supine position provides a familiar orientation of the hip joint for all or-thopaedic surgeons( Awan and Murray [2006] Khanduja and Villar [2006]), andcan be used with a standard fracture table without any special modifications. Draw-backs include difficult instrumentation manoeuvrability with obese patients, anddecreased posterior joint access (Glick and Sampson [2005]).The lateral position addresses the drawbacks of the supine approach, as thepannus in obese patients drops away from the operative field, providing better toolmanoeuvrability (Glick and Sampson [2005]), and it provides better access to theposterior and inferior joint space (Mason et al. [2003]). However, this approachrequires extra time for patient positioning and the use of special traction devicesattached to the operating table (Glick and Sampson [2005] Mason et al. [2003]).The choice of approach should defer to the surgeon?s training and habits ( Di-enst [2005]).Supine Arthroscopy Portals, Patient Position, and SetupThe supine approach, shown in Figure 1.163, is the preferred technique of ourclinical collaborator, Dr. Gilbart, and will be the focus for this thesis.In this position, the surgeon and surgical staff are on the same side as the op-erative site, with the image intensifier on the opposite side, as illustrated in 1.16.The arthroscopy video monitor is shown as placed towards the foot, however thisposition, and the number of monitors, can vary as per surgeon preference. Usingfluoroscopy, the leg is put into traction and the seal around the hip joint is releasedwith a saline-loaded needle, opening the joint capsule and creating space for thesurgical tools.3Reprinted from Operative Techniques in Sports Medicine, 13 /1, M. Dienst, Hip Arthroscopy:Technique and Anatomy, 13-23, Copyright (2005), with permission from Elsevier20Figure 1.16: Operating room setup for supine arthroscopic treatment of FAI.Dienst [2005]Three portals are primarily used for hip arthroscopy: The anterior, anterolat-eral, and posterolateral, with the first two being used primarily for the treatmentof FAI cam-type lesions on the anterosuperior neck region, which has been shownto be the most common location (Leunig et al. [2012]). The anterolateral portal isestablished first, located near the anterosuperior corner of the greater trochanter,and the anterior portal is established under arthroscopic visualization, located atthe intersection of a horizontal line running anteriorly from the tip of the greatertrochanter with a vertical line running anteriorly from the tip of the anterior supe-rior iliac spine (ASIS), as illustrated by Figure 1.174. Once the portals are estab-lished, assuming labral debridement/repair is either complete or not necessary, acapsulotomy is performed, joining the anterior and anterolateral portals to provide4Reprinted from Operative Techniques in Sports Medicine, 13 /1, M. Dienst, Hip Arthroscopy:Technique and Anatomy, 13-23, Copyright (2005), with permission from Elsevier21Figure 1.17: Placement of portals for the supine arthroscopic approach. Di-enst [2005]tools with more workspace during cam lesion treatment.The removal of the cam lesion, called an osteochondroplasty if both bone andcartilage are removed, is performed with a handheld shaver fitted with a 5.5mmspherical burr. The shaver is inserted through the anterior portal, visualized by thearthroscope in the anterolateral portal, and the margins of the lesion are mapped outusing fluoroscopy. Once the size and shape of the osseous growth are determined,they are removed using the burr to a depth that restores normal range of motion,illustrated in Figure 1.18, often 5.5mm in depth as surgeons use the diameter ofthe burr to guide resection. Once the osteochondroplasty is complete, the capsu-lotomy can be sutured at the surgeon?s discretion, the arthroscope is removed andthe portals are sutured.ComplicationsIt has been shown that the accuracy of cam lesion resection impacts the surgicaloutcome (Philippon et al. [2007]). Over-resection places the patient at a risk of22Figure 1.18: Arthroscopic view of a cam lesion (A). Spherical burr perform-ing osteoplasty (B). FH is femoral head. OST is osteoplasty. Ejnisman[2011]femoral neck fracture, although the risk is quite low with one study reporting a fre-quency of 1/120 cases (Philippon et al. [2007]). ***Another study demonstrated,using the finite element method (FEM) and cadaveric specimens, that up to 30% ofthe diameter of the femoral neck can be removed before the neck becomes the mostlikely site of fracture (Alonso-Rasgado et al. [2012]). Conversely, under-resectionof a cam lesion is far more common, likely due to the severity of a femoral neckfracture. Revision rates for FAI interventions are 4.8% (Willimon et al. [2011]),with 79-92% of revision surgeries caused by inadequate resection of a cam lesion(Philippon et al. [2007] Sampson [2005]).1.3.7 Computer-Assisted Femoroacetabular ImpingementInterventionsAs discussed in Section 1.3.6, the margins of the cam lesion are usually mappedout using the arthroscope and interoperative fluoroscopy. Adequate visualizationof the surgical site remains one of the most challenging aspects of FAI arthroscopy,and can impact the ability of the surgeon to adequately remove the cam lesion.Fortunately, technology that can interoperatively display a clear margin for boneremoval are beginning to emerge (Nawabi et al. [2013]).Arguably the first group to track arthroscopic tools interoperatively was Mon-ahan and Shimada [2008], who used an encoder linkage system which was cali-23Figure 1.19: A patient with a normal alpha angle less than 50? (A) and apatient with a cam lesion at the femoral head-neck region, with alphaangle measuring 65? (B). Dimmick et al. [2013]brated preoperatively using computer-aided design (CAD) data from a 3D modelof the hip, which was later 3D printed and used in their study. They showed thatusing the computer assisted surgery (CAS) system resulted in a 38% reduction intask completion time and 78% reduction in tool path length.Brunner et al. [2009] performed a prospective study that randomized partici-pants into navigated and non-navigated arthroscopic cam lesion treatment groups.The study used the alpha angle, as shown in Figure 1.195 to determine the amountof resection in the navigated group. The alpha angle is formed by a line drawnparallel to the femoral neck axis, from a superior perspective, and a line from thecenter of the femoral head to the transition of femoral head into the femoral neck.They were unable to show any significant effects of navigation on the clinicaloutcomes between patients with sufficient and insufficient correction of the alphaangle, highlighting the limitations of the alpha angle as an outcome measure, andthe importance of accurate preoperative planning. Most recently, a group (Eckeret al. [2012]) using a heavily modified version of the software from Brunner et al.[2009], showed that the discrepancy between planned and actual reaming was lessthan 1mm in 18 operations, performed by two surgeons.5Reprinted from Radiologic Clinics of North America, 51/3, Dimmick et al., FemoroacetabularImpingement, 337-352, Copyright (2013), with permission from Elsevier241.3.8 Motivations for Displaying a Clear Boundary for BoneRemovalCurrent CAS systems for hip arthroscopy have been shown to reduce simulatedprocedure time but it remains unclear if they improve accuracy of resection, par-ticularly in live arthroscopic surgeries to treat FAI. However, the positive effectsof displaying a clear margin for resection of bone has been demonstrated in or-thopaedic procedures for other joints.Arthroscopic bone shaping tools controlled completely by the surgeon are doneso with a technique called freehand, in a manner similar to procedures in whichthe tool position is not constrained by a jig. Conventional orthopaedic freehandtool use in combination with CAS navigation systems has been shown to reducecutting time by 15% and decrease implant translational malalignment by 70-90%in a phantom model for minimally invasive total knee arthroplasty (TKA) (Haideret al. [2007]). Clinical data for conventional freehand TKA produces similar bene-fits, enabling better implant alignment, standing alignment, and reduced blood loss(Chauhan et al. [2004]). Studies demonstrate combining freehand tool use withcomputer generated margins for resection results in improved accuracy and poten-tial for reduced procedure time, although the orthopaedic field has moved towardsnavigated cutting guides as generally described in Kendoff et al. [2010] which of-fer higher precision compared to freehand. Unfortunately, navigated cutting jigsare not compatible with an arthroscopic approach, so they cannot be used for thetreatment of FAI. We are therefore interested in determining how much additionalbenefit is possible in freehand navigated FAI procedures if bracing is also imple-mented.Freehand bone milling, the shaping of bone using a burr rather than an os-cillating blade, in combination with CAS remains a relatively new approach, asexemplified by the NavioPFS (Brisson et al. [2004] Smith et al. [2013]) for uni-compartmental knee arthroplasty (UKA) procedures. It uses a freehand millingtool in combination with a navigation system that displays the boundaries for boneremoval, with a burr that can stop rotating or retract inside the tool to prevent over-resection of bone. However, the NavioPFS remains a very expensive option whencompared with traditional approaches, and the benefits of this supplementary me-chanical approach in addition to CAS navigation for freehand milling has not yet25been demonstrated. Further studies should investigate what the incremental bene-fit of bracing might be, assuming it is used in parallel with a ?perfect? navigationsystem which could be implemented by providing a colour-based contrast betweenthe over and under-resection regions.1.4 Summary and Research GoalsLiterature has shown that arthroscopic treatment of FAI is a technically demandingprocedure, with accurate removal of a cam lesion often dictating the success ofsurgical intervention. In section 1.1 literature showed that bracing offered a low-cost, intuitive way to improve precision in humans and robots alike. The goalof this study was to design a bracing strategy that enables the user to be betterand faster at performing a simulated femoral head-neck osteochondroplasty millingthan currently employed orthopaedic techniques.1.4.1 Hypotheses1. Bracing will reduce the length of time taken to remove a phantom cam lesioncompared to traditional arthroscopy.2. Bracing will reduce the error associated with removing a phantom cam lesionin a traditional arthroscopic manner.1.5 Thesis OrganizationChapter 1 provides an overview of relevant literature in bracing and arthroscopictreatment of FAI and details the hypothesis of this thesis.Chapter 2 details a preliminary study into the feasibility of applying bracingto a 3DOF milling task focused on the treatment of cam-type femoroacetabularimpingement.Chapter 3 elaborates on the pilot study performed in Chapter 2 by presentinga targeted redesign of the tool brace in parallel with the development of a higherfidelity surgical simulation.Chapter 4 details our study of the brace and surgical simulator from Chapter 3on a cohort of 20 non-surgeons and 5 experienced arthroscopic surgeons.26Chapter 5 concludes with a discussion on our findings, their relevance to realorthopaedic procedures, and considerations for future work in the area of bracedorthopaedic surgery.27Chapter 2Preliminary Exploration ofBracingContent of this chapter was presented at the 13th meeting of the International So-ciety for Computer Assisted Orthopaedic Surgery (CAOS) (Kooyman and Hodgson[2013b]) and published in condensed abstract form in The Bone & Joint Journal(Kooyman and Hodgson [2013a]), formerly known as JBJS(Br).2.1 Applying Bracing to a Current Orthopaedic ProblemBracing, as discussed in depth in section 1.1, is a strategy employed by humans androbotic devices to improve performance in a variety of tasks such as writing andautomated grinding (Book et al. [1984]). It can be more generally described as aparallel mechanical link between the actor, the environment, and/or the workpiecethat alters the mechanical impedance between the tool and workpiece in order toimprove task performance, illustrated by figure 1.2.As an extension of recent 1 DOF work in UBC?s Neuromotor Control Labora-tory, we investigated the potential value of bracing in the context of the surgicallyrelevant 3 DOF task of bone milling with a spherical burr. More specifically, wesimulated the milling of cam-type FAI lesions, as described by Ganz et al. [2003].This procedure was chosen due to the growing recognition of FAI as a mechanismfor the development of early OA (Philippon et al. [2007]).28Figure 2.1: A visual overview of the stages in the pilot study.2.2 Research ObjectiveAs demand for surgical intervention increases, surgeons are reluctant to embracetechnologies that require increased procedure time. We hypothesize that bracingcan enhance precision in simulated femoral head-neck osteochondroplasty withoutextending procedure time. The goal of this study was to evaluate whether a pro-posed bracing technique could enable a user to perform a resection on a femoralmodel more accurately and quickly than a currently employed arthroscopic tech-nique (Byrd [2005]).2.3 MethodsBriefly, illustrated in Figure 2.1, proximal femur segments were cast from urethaneplastic, detailed in 2.3.1, and then scanned to obtain ground-truth surface data, de-tailed in 2.3.2. Then, cam lesions were cast onto the anterosuperior neck region,detailed in 2.3.3, and removed during a mock surgery, detailed in 2.3.4. The prox-imal femur segments were laser scanned again to obtain surface information fromthe surgery, and a deviation analysis was performed using Rapidform software,detailed in 2.3.5.29Figure 2.2: A silicone mold (A) was created from the proximal segment of acommercially available adult left femur. 16 Proximal femur segmentswere cast in this mold (B) using white urethane plastic2.3.1 Proximal Femur CastingProximal femur segments were cast using a silicone mold (Oomoo R?30, Smooth-on Inc.) of a commercially available adult left femur (Sawbones R?, Pacific Re-search Laboratories). The silicone mold was filled with white urethane plastic(Smooth-Cast R?300, Smooth-on Inc.). This casting process was adapted fromSimpson [2010], who used Oomoo R?30 and Alumilite casting resin to make ver-tebral segments. The Shore D hardness of the urethane plastic is reported as 70D,similar to human cortical bone (Kincaid et al. [2007]), indicating that the urethaneplastic should provide similar milling properties. Following casting, the meanweight of the femurs was 233.62 +/- 2.77g, or less than 1% standard deviationin weight. Laser scanning analysis, detailed in section 2.3.2, revealed that thebetween-femur deviation, or the average deviation in the surface profile, was 0.018+/- 0.006mm. We conclude that the casting process was sufficiently repeatable forour purposes as it was below our scan-rescan upper 95% error boundary of 0.037,detailed in B.1.2.3.2 Laser ScanningA VIVID 9i 3D laser scanner, specifications in Appendix A, was used to digitizethe samples. Scans were performed in a dim room, taken at 30? rotations aboutthe femoral neck axis to maximize visibility around the region of interest (ROI).30Samples were placed on a rotating stage and held using plasticine, allowing thesamples to be placed in approximately the same position during scanning. Thisscanning protocol was refined with input from Rapidform engineers1, who sug-gested switching from 60? to 30? scan intervals, in addition to removing as muchlight, both natural and artificial, as possible to improve scan quality.2.3.3 Cam Lesion Casting3D models of proximal femurs with cam lesion morphology, provided by a col-league, were viewed and a representative cam lesion was created, based on theestimated size and shape of the lesions segmented from clinical data. Ethics per-missions prevented quantitative measurements of cam lesion dimensions from be-ing taken.A negative mold was created from this representative cam lesion, and a twopart casting system, shown in Figure 2.3, allowed the lesion to be cast directlyonto the surface of the proximal femur segment. Cam lesions were cast usingblack urethane plastic(Alumilite Regular, Alumilite Corporation) to provide a clearboundary to the white proximal femur to guide accurate resection during surgery,acting as a CAS system which is able to display a perfect boundary between overand under-resection, the benefits of which are discussed in Section 1.3.8 . TheShore D hardness of the urethane plastic was reported as 75D, again providingsimilar hardness properties to human cortical bone (Kincaid et al. [2007]). Litera-ture specifically addressing the comparative machinability of this urethane plasticand bone was not found. However, one might expect a difference based on thestructure of the two materials, the urethane offering an isotropic structure, whereasthe bone would be anisotropic and influenced by the osteons. The mean weight ofthe cam lesions was 5.31 +/- 0.20g, or less than 4% sample-to-sample variation.Milling time depends most strongly on volume, so we measured and controlled forthe mass of the samples. It was decided, due to the sub-gram sample variations,that the two-part casting process was suitable for the study.1I Cullimore (personal communication, November 8 2012)31Figure 2.3: A silicone mold showing a cam lesion created by pouring blackurethane plastic in from a hole in the exterior of the mold, providing aperfect fit to the surface of the proximal femur segment.2.3.4 Mock SurgeryTwo binary factors were tested, resulting in four test conditions:1. Braced vs. Unbraced: In the unbraced case, the tool was held using a gripapproximating that used in current surgeries; in the braced case, the tool wassupported using a spherical bearing mounted in the approximate anterolateralarthroscopic portal position, with an insertion depth of approximately 5cm(measured from the insertion point of the bearing to the closest edge of thecam lesion), as illustrated in Figure 2.5. The brace provided no axial support,and had a conical workspace of approximately 135?.2. Speed vs. Accuracy: The subject was instructed either to perform the resec-tion as quickly as possible with moderate regard for accuracy or as accuratelyas possible with moderate regard for time.32Figure 2.4: Cam lesions cast from black urethane plastic. It can be seen that alearning curve is present in the two-stage casting method, however onlyappropriate samples such as the far right were used in experimentation.Pilot test subjects were 4 adult males (25 +/- 3 years) with no surgical ex-perience. ***Subjects were selected based on initial selection criteria of normalor corrected-to-normal vision and have no history of neuromuscular injury to theupper body. Additional selection criteria was not included as the study author be-lieved the mock surgical task to be sufficiently simple that an adult could performit successfully without specialized training. Each subject performed one resectionunder each test condition in a randomized order, for a total of 16 samples, us-ing a Dyonics shaver handpiece and 5.5mm Acromioblaster shaver burr (Smith &Nephew Advanced Surgical Devices). Prior to performing surgery, subjects weregiven 5 minutes to use the shaver on an identical phantom to the one they weretested on, however they were not given an opportunity to practice with the brace.Subjects were timed by analysing video footage post-hoc, with time starting whenthe tooltip touched the lesion, and ending when the entire lesion had been removed.Task posture was not constrained as it was desired to observe the variations be-33Figure 2.5: The mock surgical setup for simulated femoral head-neck osteo-chondroplasty. Subjects would use a handheld arthroscopy shaver with5.5mm spherical burr to remove the black cam lesion with or withoutthe aid of the spherical bearing tool brace, placed in the approximateposition of the anterolateral portal.tween subjects.Following the removal of the lesion, femurs were laser scanned as described inSection 2.3.2 to acquire the post-resection surface geometry.2.3.5 Rapidform XOR2/XOV2 AnalysisPre- and post-resection laser scans were transformed into 3D models using Rapid-form XOR2 (3D Systems Corporation), as shown in Figure 2.6, with the pre-resection model acting as the control for mesh comparisons. Post-scanning refine-ments were kept to a minimum to prevent loss of surface detail. The 3D modelswere registered to one another and then trimmed to the anterior side of the femoralhead-neck region where the lesion was located, as shown in Figure 2.7, to eliminateextraneous information. The trimmed models were then loaded into Rapidform34Figure 2.6: Mesh-rebuild process in Rapidform XOR2.XOV2 (3D Systems Corporation) to inspect the meshes, and a whole-mesh devi-ation analysis was performed by comparing the pre- and post-resection models toone another, as illustrated in Figure 2.8. The deviation was calculated by project-ing the scanned points (post-resection) onto the nominal (pre-resection) data anddetermining the gap between the two. The results are reported as RMS deviationsover the trimmed surface, with an RMS error surface map shown in Figure 2.9,matched with the timed procedure length.2.4 ResultsThe effects of bracing on time and accuracy, with the speed condition, are shownin Figure 2.10. Subjects 1 and 2 both experienced effects of bracing that reducedprocedure length and RMS error. Subject 3 reduced their procedure length at theexpense of an increase in RMS error when braced, and Subject 4 experienced smallnegative effects of bracing on both outcome measures.The accuracy condition results are shown in Figure 2.11. Subjects 1 and 2both experienced effects of bracing that reduced procedure length and RMS error.35Figure 2.7: 3d models of the proximal femur segments were trimmed to theROI to prevent influence of extraneous information.Figure 2.8: The RMS error in Rapidform XOV2 is determined by projectingthe post-resection data onto the pre-resection data, then calculated thedifference between the two surfaces via a projection to the closest point.36Figure 2.9: A visual representation of the RMS error in Rapidform XOV2.Thinking of the colours as a depth map, the positive effects of bracingare evident.37Figure 2.10: Results from the simulated femoral head-neck osteochon-droplasty, with subjects being told to be as fast as possible with moder-ate respect for accuracy. RMS error (mm) is on the x-axis, and proce-dure length (s) is on the y-axis. Hollow shapes represent the unbracedcondition, and filled shapes represent the braced condition.Subjects 3 and 4 both reduced their procedure length at the expense of increasedRMS error under bracing.The combined effects of bracing on time and accuracy of femoral head-neckosteochondroplasty are shown in Figure 2.12 and summarized in Table 2.1, withdata presented as the mean and standard deviation of the 4 subjects. As expected,when asked to emphasize accuracy over speed, accuracy improved and time in-creased under both bracing conditions. In both accuracy and speed cases, bracingtended to reduce errors (on the order of 7-14%) and task duration (on the order of32-52%), although, given the small number of subjects in this pilot study, thesedifferences were not statistically significant.38Figure 2.11: Results from the simulated femoral head-neck osteochon-droplasty, with subjects being told to be as accurate as possible withmoderate respect for time. RMS error (mm) is on the x-axis, and proce-dure length (s) is on the y-axis. Hollow shapes represent the unbracedcondition, and filled shapes represent the braced condition.Table 2.1: Combined results from the pilot study.Test Condition Procedure Length (s) RMS Error (mm)Unbraced Speed 305 ? 113.5 0.477 ? 0.123Unbraced Accuracy 496 ? 125.4 0.339 ? 0.154Braced Speed 201 ? 87.4 0.445 ? 0.125Braced Accuracy 377 ? 179.6 0.298 ? 0.05339Figure 2.12: Combined results from the simulated femoral head-neck osteo-chondroplasty. RMS error (mm) is on the x-axis, and procedure length(s) is on the y-axis. Hollow shapes represent the unbraced condition,and filled shapes represent the braced condition. Data is presented asthe mean +/- standard deviation.2.5 DiscussionIn this pilot study, the results shown in 2.4 provide some encouragement that ourhypothesis that bracing can improve both speed and accuracy of cam lesion resec-tion by untrained subjects may be true. The standard deviations between subjectsare high and are likely due to both the difficulty of the task and differences in ex-perience using handheld power tools, so additional subjects would be needed toobtain sufficient statistical power to verify the trends identified here.The trend of bracing reducing both procedure length and error can be thoughtof in a more general medical context, where most existing surgical-assistance tech-nologies are thought to exist on an isoline that trades off time or error, illustrated inFigure 2.13. Robotic technology typically offers very low error at the expense ofprocedure length. Navigation technology can improve the accuracy of traditional40Figure 2.13: A hypothetical categorization of existing orthopaedic surgicaltechnology on a time-error plane with the ideal scenario of zero timeand zero error interventions displayed in the bottom left corner.cutting jigs/guides again, at the expense of procedure length. More recent advancesin freehand navigation show potential to improve on both time and accuracy. Webelieve that bracing holds potential for offering significant improvements in bothtime and accuracy.2.5.1 Study LimitationsActing as a preliminary exploration into the effects of bracing, there are severallimitations present in this study.It is currently unclear as to why some subjects experienced paradoxical effectsof bracing, shown in Figures 2.10 and 2.11, where bracing caused a negative in-crease in one or both of the time and error outcome measures. This could be due toeffects of learning or prior power tool experience impacting their ability to inter-nally map the effects of the brace, although testing order was randomized to prevent41Figure 2.14: Three of the four subjects used in the study, each displayingdifferent task postures. (A) is bracing against the workpiece, (B) isbracing against their trunk, and (C) is bracing against the target byshortening the effective tool tip length.Figure 2.15: The spherical bearing tool brace did not provide sufficientworkspace to allow for the entire cam lesion to be removed.a uniform impact. Alternatively, each subject adopted a different task posture dur-ing the unbraced configurations, coincidentally employing some form of bracingeither against themselves or the workpiece, shown in Figure 2.14. The simplebrace used in this study would not be directly suitable for use in live surgical pro-cedures as it greatly reduced the workspace of the arthroscopy shaver and thereby,in combination with the shape of the greater trochanter, prevented full resectionof the lesion; this workspace limitation potentially led to an underestimate of thevalue of bracing in this study. In addition, other studies have shown that reversedmotions due to operating across a fulcrum point, as was done here, can impair per-42formance in laparoscopic tasks (Hodgson et al. [1999]), and we hypothesize thatperformance in a braced configuration could be further improved by altering thebrace design to prevent motion reversal.2.6 Conclusions/Future WorkThis study has shown that bracing could potentially improve task performance, al-though the results here show only a modest effect size in comparison to the control,which could limit the clinical utility. However, it is possible that different brace de-signs could produce greater effects, so we believe that a more complete explorationof the design space is warranted.In future work, we intend to focus on designing and fabricating a more effectivebrace for the targeted surgical task that will provide a more acceptable workspaceand will have greater potential to be translated to the operating environment. Wealso intend to test subsequent designs with a larger pool of subjects, in order to testwhether the performance improvements observed here can be achieved reliablyacross a more representative user population.43Chapter 3Bracing Apparatus DesignContent in this chapter prepared for submission to Clinical Orthopaedics and Re-lated Research for publication.Several problems with the study in Chapter 2, highlighted in Section 2.5.1,were associated with the design of the fulcrum-brace. Workspace limitations pre-vented full resection of the cam lesion and it has been shown that reversed mo-tions due to operating across a fulcrum point can impair performance in laparo-scopic tasks Hodgson et al. [1999]. We believe that these two issues may haveled to an under-representation of the potential benefits of tool bracing in simulatedfemoral head-neck osteochondroplasty. In order to address these possibilities, wereassessed the design of the brace.3.1 Testing Apparatus RedesignThe surgical simulation performed in Section 2.3.4, while suitable for a preliminaryexploration, lacked the desired fidelity for adequate surgical simulation. The useof binocular, direct vision instead of a 30? arthroscope on a two dimensional (2D)monitor eliminated the difficulties associated with triangulating tools in three di-mensional (3D) space, adding in heightened depth perception and the ability to seethe target from different angles, an impossible feat in real procedures. Further,it was confirmed by our clinical collaborator that the arthroscopic surgical instru-ments receive support from the soft tissues during the procedure, altering how they44Figure 3.1: Arthroscopes display their field of view at an angle to the axisof the scope, denoted by ? . Typical arthroscopes used in FAI interven-tions are 30? and 70?, although the latter is used predominantly duringacetabular procedures.perform in vivo. Finally, casting the proximal femur segments from 2.3.1 took ap-proximately 30 minutes per sample, which could make larger studies prohibitive.A test sample that allowed for rapid production was desired.3.1.1 Arthroscope AdaptationWe addressed the vision issue by switching to an arthroscope proxy, which con-sisted of a flexible USB borescope inserted into a metal tube that allowed manipu-lation with the non-dominant hand in a manner similar to a real arthroscope. Theproper medical device was not used as the risk of an unskilled subject damaging theoptics during testing and the replacement cost was too high. The handheld USBborescope has a straight-on field of view, uses a ring of LEDs to illuminate theviewing area, and has a 1cm focal length. For comparison, a standard 30? arthro-scope that would be used during the milling component of FAI surgery would havea field of view at a 30? angle to the length of the tool, illustrated in 3.1, use fibre op-tics in combination with a xenon light source, and has a focal length of 30-40mm.The view from a standard and simulated arthroscope is shown in Figure 3.2 andwas judged to be of sufficient similarity for our simulation, and would later beconfirmed by a surgeon population in Chapter4.3.1.2 Required Fidelity for SimulationThe variations observed in task posture from Section 2.5.1 was an indication thatmore attention to the surgical simulation was needed. Subjects were adopting task45Figure 3.2: A comparison of a 30? arthroscope (A, C) against our proposedUSB adaptation (B, D), showing an unresected sample (A, B) and aresected sample (C,D).postures that were not necessarily suitable in a live surgery, and allowing uncon-strained posture would impact the validity of future work. In order to determineif a bracing solution was superior to an unbraced configuration, we would need tomimic the surgical environment as closely as possible, within reason.Through surgical observation and discussion with our clinical collaborator, welearned that the mechanical interactions between the soft tissues and the arthro-scopic tools affected their handling properties by altering the available workspaceand the stiffness in varying directions. We also learned that the task posture wasconstrained to milling through the anterior portal using a hammer-type grip, toallow for interoperative fluoroscopy using a standard AP view.We decided that the most important characteristics for surgical simulation were:1. Task Posture - Restricted to that found in the supine position during millingthrough the anterior portal.2. Tissue Stiffness/Impedance - A reasonable proxy for the support provided to46the tool by the patient?s soft tissues (musculature and joint capsule) must beincluded as they impact the handling properties of the tools.3. Cutting Behaviour - Using blades designed to cut bone suggests that theyare optimized for a certain material hardness. The material used for the camlesions should mimic the hardness properties of bone for somewhat realisticmachining characteristics.By including the above characteristics, we believe that the most important me-chanical properties affecting the tool handling and milling attributes are accountedfor. Other aspects of the simulation that we believe are less important are:1. Fluids - During arthroscopy, natural and irrigation fluids may alter the abilityof the subject to hold the tool, as well as the ease of insertion of the tools intothe arthroscopic portals. However, interoperatively the tool would be cleanedif it became slippery, and the insertion of the tool into the portal is not partof the milling procedure.2. Temperature - We felt that the natural body temperature would not impactthe fidelity of the simulation in any meaningful manner.3. Surface Texture - The surface texture of the bone, influenced by the porosityof the cortical structure, was believed to have minimal influence on the grossmilling properties that we are interested in simulating.3.1.3 Soft Tissue Interface Design and IntegrationA soft tissue interface was designed using measurements taken from a porcinemodel, detailed in Appendix C. We also observed several FAI procedures per-formed by our clinical collaborator, paying attention to the portals used to gainaccess to the anterosuperior region of the joint during the milling portion of theprocedure and the insertion depth of the tools. Contrary to the approach used inSection 2.3.4 where milling was performed through a simulated anterolateral por-tal, our clinical collaborator preferred to insert the arthroscope through the antero-lateral portal and insert the milling tool through the anterior portal. The portals, asdescribed, are illustrated in Figure 1.17.47Figure 3.3: A hip/femur model was inserted into a mannequin and coveredwith a foam surface to allow for palpation of bony landmarks used totriangulate arthroscopic portal locations.The surgical approach was reproduced using a Sawbone hip and femur modelinserted into a mannequin torso, donated by colleague Ryan Payne, shown in Fig-ure 3.3. The surgical site was covered with a blue foam which allowed bony land-marks, required for locating the portals used in surgery, to be palpated. From thissetup, the approximate approach angle of the shaver tooltip relative to the plane ofthe femur was measured to be ? 70?.Discussion with our clinical collaborator also revealed that the support givento the tool was anisotropic due to a capsulotomy that is normally performed; a cutmade to join the anterior and anterolateral portals. This would allow the tool tomove in a relatively uninhibited manner parallel to the line joining the portals butwould greatly hinder movement perpendicular to the cut. We felt it was importantto include this effect in our surgical simulation.In parallel with Section 3.1.4, a new simulation box was designed incorporatingthe USB arthroscope from Section 3.1.1, the soft tissue proxy from Appendix C,the revised surgical approach detailed above, and the capsulotomy, and is illustratedin Figure 3.4. The distance from the foam-based joint capsule to the target wasbased on arthroscopic views of the milling tool during surgery, such as the viewfrom Figure 1.18B where the length of the blade guard was used to obtain theapproximate distance from the joint capsule to the target area. The thickness of thefoam was based on observation of our clinical collaborator during surgery thoughthe insertion depth, and subsequently the amount of support the tool receives, is48Figure 3.4: An overview of the soft tissue simulation used in the present ex-periment. 2050 foam and 5mm thick closed-cell foam (A) were placedin a wood box, forming the STS condition, which housed the 3D printedsample holder (B). The capsulotomy, which joined the tool and cameraportals, is shown as a slit in (B). The distance from the capsule to thesample, shown in (C) was based off of the approximate insertion depthof the tool as observed during surgery. The final box with soft tissueinterface, is shown in (D) with the fulcrum brace in place to highlightthe anterior portal.clearly patient-dependent.3.1.4 Testing Sample RedesignIn order to gather more data in a reasonable amount of time, a redesign of theproximal femur samples from Section 2.3.1 was needed.It was evident that our primary interest was during the last phase of the millingprocess, where only a small amount of lesion had to be removed. This was whereover-resection was most likely to occur, so a small dome shape was recreated using49Figure 3.5: Mock cam lesion used in the present study. Black plastic shapeglued onto wood substrate.a silicone casting process identical to Section 2.3.1 and these pieces were thenglued onto a poplar substrate.The final model consisted of 1.5x1.5? poplar wafers with 19mm diameter,3.6mm thickness black, domed cam lesions (0.6 +/- 0.05 g) glued in the centre,as shown in Figure 3.5. The contrast between the black cam lesion and light colourof the wood provided a clear boundary for resection. The Shore D hardness of theurethane plastic was reported as 75D providing similar properties to human corticalbone (75-85D) as determined by Kincaid et al. [2007]. Laser scanning could notbe used to determine surface variations of the samples as the VIVID 9i is unable todetect the black resin. Instead, visual comparisons of the lesions were performedafter they had been glued onto the wood substrate. Since they had all been castfrom the same initial form, in the same mould, in a stable environment, obviousvariations were absent.3.2 Exploration of Bracing Design SpaceWe used our higher-fidelity surgical simulator designed to explore the bracing de-sign space in a more targeted manner. We designed four different types of braces:3.2.1 Fulcrum BraceTo maintain continuity with the brace used in Chapter 2, another fulcrum brace wasdesigned using similar parts to the previous iteration.50Figure 3.6: The fulcrum brace, consisting of a spherical bearing rod-end, wasmounted at the approximate location of the tool portal for the presentstudy.A spherical bearing rod-end was mounted at the hypothetical anterior portal,shown in Figure 3.6. Prior to testing, this brace was determined to be able to allowaccess to the entire workspace needed for the simulated task of removing the camlesion with a handheld milling tool.3.2.2 Armrest BraceFrequent reference to the self-bracing scenario of palm rest during writing, shownin Figure 1.1, made the idea of an armrest to assist during surgery appealing. Wediscussed the concept with our clinical collaborator and he confirmed that it mighthave merit as he experiences shoulder discomfort from lengthy procedures wherehe has to abduct his arm to use surgical instruments.An armrest that would support the forearm, largely isolating tool movement51Figure 3.7: The armrest, which kept the subject?s elbow at approximately 90?flexion, used in the present study.to ab/adduction of the wrist, was designed to allow the user to rest their arm at anatural height for the procedure, rather than require any biasing forces to maintaincontact. The brace design is shown in Figure 3.7.3.2.3 Remote Centre of Motion BraceMotion reversal caused by operating across a fulcrum, as performed in Section2.3.4 and required by the brace in Section 3.2.1, has been shown to impair perfor-mance during laparoscopic tasks (e.g. see Hodgson et al. [1999]).In order to correct for this motion reversal, a remote centre of motion (RCM)was placed below the target area, following a projection of the access point pro-vided by the fulcrum brace in Section 3.2.1, as shown in Figure 3.8. This elimi-52nated the motion reversal by mechanically requiring that the tool rotate around thelocation of the spherical bearing beneath the workspace.3.2.4 Remote Centre of Motion Brace with Axial StiffnessPrevious work performed by Jake McIvor has shown that 1DOF bracing can reducedrill plunge-through. This intentional alteration of the axial stiffness was the moti-vation behind the alteration of the brace design in Section 3.2.3. Introducing axialstiffness could reduce the amount of over-resection during the simulated surgicalprocedure.A spring with stiffness of 1700 N/m was placed such that when the tool wasmoved towards the target area it would be compressed, adding axial stiffness,shown in Figure 3.9. This spring was selected based on dimensions and avail-ability, and would later be optimized for the task (see D) Rotation about the RCMcompressed the spring a negligible amount, so the added stiffness was predomi-nantly restricted to the axis of the tool. The zero-point of the spring was set sothat the tip of the milling tool had just penetrated the simulated joint capsule, andfurther axial movement involved compression of the spring.3.3 Brace Design TestingWe performed a small study to determine the optimal brace design to bring forwardto a larger population user study.3.3.1 MethodsSubjects were tested in a randomized order in the following test conditions:1. Unbraced - No tool support was available and direct vision was blockedusing a black lycra covering over the test box.2. Soft Tissue Support - No tool brace was used. Tool support was provided bythe soft tissue interface, serving to simulate performance in actual surgery.3. Fulcrum Brace - The brace detailed in Section 3.2.1 was used in addition tothe soft tissue interface.53Figure 3.8: The RCM was placed in a location that projected the instantaneouscentre of rotation along the tool insertion vector required by the surgicalapproach and the fulcrum brace, at a depth of 2X the insertion depth.54Figure 3.9: The RCM brace from Figure 3.8 shown here with a spring thatadded axial stiffness to the tool.Figure 3.10: (A) A closeup of the RCM brace from Section 3.2.3, and (B) theRCM + Spring brace from Section 3.2.4, highlighting the differencebetween the two designs.554. Armrest Brace - The brace detailed in Section 3.2.2 was used in addition tothe soft tissue interface.5. RCM Brace - The brace detailed in Section 3.2.3 was used in addition to thesoft tissue interface.6. RCM + Spring Brace - The brace detailed in Section 3.2.4 was used in addi-tion to the soft tissue interface.Test subjects were 4 adult males (23.5 +/- 2.6 years) with no surgical experi-ence. Each subject performed three resections under each test condition in a ran-domized order, with the exception of the RCM + Spring condition where only twosubjects participated, for a total of 15-18 samples, using a Dyonics shaver hand-piece and 5.5mm Acromioblaster shaver burr (Smith & Nephew Advanced Surgi-cal Devices). Prior to performing surgery, subjects were given one sample to usethe shaver on as practice to become familiar with that particular brace condition.Subjects were timed by an observer, with time starting when the tooltip touchedthe lesion, and ending when the entire lesion had been removed. Task posturewas constrained to one that mimicked the approach used by our clinical collabo-rator; Subjects held the shaver with a downwards hammer grip in their dominanthand, briefly illustrated in Figure 3.7, with the USB arthroscope held in their non-dominant hand.Laser ScanningFollowing the removal of the lesion, samples were prepared for laser scanningby receiving a light coat of white spray paint. This enabled the laser scanner todetect any unresected lesion, as illustrated in 3.11. This light coating of paint alsoenabled the laser scanner to detect the surface of the wood by reducing the porosity,improving the quality of the scan.Laser scanning was performed as described in section 2.3.2, with a VIVID 9i3d laser scanner, with a wide angle lens at a distance of 600mm to the sample.Samples were placed in a custom jig, shown in Figure 3.12, that provided repeat-able positioning relative to the scanner, ensuring that all samples were scanned inan identical manner. Scans were taken at 30? intervals, ranging from +/- 60? to56Figure 3.11: Small amounts of black lesion, under 1 mm2 as visually assessedby the facilitator, can remain after the mock surgery provided the sub-ject feels that they have adequately resected the cam lesion. In the toptwo cases, the subject would be allowed to continue the resection. Inthe bottom two cases, the resection would be considered complete.ensure that only the front portion of the sample was captured. This reduced thescanning time by approximately 50%, compared to scanning the front and back ofthe sample.Rapidform XOR2/XOV3 AnalysisPost-resection laser scans were transformed into 3D models using Rapidform XOR2(3D Systems Corporation). Scan data, acquired using the methods outlined inSection 3.3.1, was loaded into Rapidform XOR2 (Figure 3.13 A) using the meshbuildup feature, aligned using a global reference iterative closest point feature (Fig-ure 3.13 B), merged into a single model (Figure 3.13 C), and finalized by redis-tributing the poly faces (Figure 3.13 D). The final model, shown in detail in Figure3.14 A, had the edges trimmed off, Figure 3.14 B, to eliminate extraneous informa-57Figure 3.12: A custom 3D printed sample holder was used to keep each sam-ple in a repeatable position during laser scanning.tion during mesh deviation analysis. Rather than a pre-resection scan, as per 2.3.2,a reference plane was fit to the planar surface of the wood substrate by picking sev-eral points on the surface unaffected by the lesion removal. A repeatability analysisof the reference plane fitting feature, detailed in Section B.3, found that two stan-dard deviations, representing the variability of 95% of the data, was 0.021mm,which is less than the upper 95% error bound reported in the scan-rescan analysisfrom Section B.1. As such, differences greater than 0.04mm are likely to be real.A mesh-deviation calculation from this reference plane was performed, illus-trated in Figures 3.16 and 3.17. The results are reported as root mean square (RMS)deviations, shown in Figure 3.18, over the trimmed surface.3.3.2 ResultsThe results of the testing outlined in Section 3.3 are summarized in Figure 3.18.The RCM+Spring category had only two subjects compared to 4 subjects for theremainder of the bracing conditions.Bracing of all types reduced procedure length and RMS error between 40-50%and 30-40% respectively, moving towards an ideal performance taking zero sec-58Figure 3.13: Laser scans are trimmed of extraneous data (A) and aligned (B-C) and merged into a final mesh (D).Figure 3.14: The edges of each sample (A) are trimmed (B) in RapidformXOR2 to reduce extraneous data and improve the fit of the referenceplane and resulting mesh deviation calculation.59Figure 3.15: A reference plane (A) is fit to a sample by selecting severalpoints on the surface of the sample unaffected by the cam lesion re-section. A side view is shown in (B).onds and having zero error shown in the bottom left corner of Figure 3.18. Themost effective brace was the RCM+Spring, described in 3.2.4, and the least effec-tive was the RCM.StatisticsThe statistical analysis package, IBM c?SPSS R?20, was used to analyse the resultssummarized in Figure 3.18. The Shapiro-Wilk test was used to assess the normal-ity of the data, and was significant for both procedure length (p < 0.0000001) andRMS error (p < 0.000028), indicating that the data was not normally distributed.A log-normal transform of the data was performed in an attempt to obtain nor-mally distributed data. The Shapiro-Wilk test returned significant results for bothLog(Procedure Length) (p < 0.015) and Log(RMS error) (p < 0.011), indicatingthat non-parametric tests must be used.The non-log-normal data was then rank-transformed, with the smallest valuebeing given rank 1.The Kruskal-Wallis test for independent samples detected a significant effect of60Figure 3.16: The RMS error in Rapidform XOV3, using the reference planemethod, is determined by projecting the post-resection data onto thereference plane data, then performing the calculation based on themesh deviation error.bracing on Procedure length (p < 0.002) and RMS error (p < 0.002), which allowsus to reject the null hypothesis that the rank distribution of our variables was thesame across all levels of bracing.The Mann-Whitney U Test for pairwise comparisons was used with a Bonferroni-Holm correction for multiple comparisons to determine which bracing conditionsproduced significant effects on procedure length and RMS error, with results, re-ported as the two-tailed p-value, summarized in Table 3.1. There was a significanteffect of bracing on procedure length between the unbraced and STS (p < 0.001),unbraced and armrest (p < 0.0004), and unbraced and RCM+spring (p < 0.004)conditions. There was a significant effect of bracing on RMS error in the unbracedand STS (p < 0.002) and the unbraced and RCM+spring (p < 0.004) conditions.As we desire a bracing solution that improves the performance of subjects ina surgically-relevant environment, a power analysis was performed between theSTS and RCM+spring conditions to determine the required sample size to detect adifference, with the same magnitude as the one in this study, if one existed. Based61Figure 3.17: Two examples of the mesh deviation visualization from Rapid-form XOV3. The deeper the blue, the greater the amount of over re-section that has occurred.on the means from both procedure length and RMS error, the power analysis showedthat 54 samples in each group would be required for a power of 0.8, indicating thatat least 18 subjects, performing 3 repetitions each, would be required in any futurestudy to detect a difference in either or both procedure length and RMS error.3.3.3 DiscussionIn this study we have built on the results from Chapter 2 by designing an im-proved surgical simulation and a wide range of bracing types. The results providedhere, where there was a statistically significant effect of bracing on RMS error andprocedure length for the STS and RCM + spring braces compared to the unbracedcondition, continue to provide encouragement that our hypothesis that bracing canimprove both speed and accuracy of cam lesion resection by untrained subjectsmay be true. The standard deviations between subjects remain high, as we foundSection 2.4, and again are likely due to the difficulty of the task, particularly in theunbraced configuration, and the differences in experience using handheld tools.62Figure 3.18: Combined results showing the performance of the differentbrace designs compared to an unbraced control.The RCM + spring brace gave the best performance, however it was only tested ontwo subjects due to time constraints. However, as the design of the braces becomesmore complex, there is some possibility that the subjects may not have sufficienttime to fully understand the stiffness alterations induced by the bracing linkages,which may induce a learning period during which any benefits of bracing long termmay be less apparent. As indicated by power analysis, an increased sample size ofapproximately 18 subjects will be required to determine if the RCM + Spring braceimproves subject performance compared to the surgical simulation condition ofSTS only, if the difference is on the order of 15%, as we found in this pilot study.By intentionally controlling vision by restricting the subjects to using a USBborescope, we kept the field of view largely similar between subjects except in theunbraced case (shown compared to cases where foam is present in Figure 3.19),where subjects would draw the camera towards them to gain a larger field of view.This might be due to the relatively long focal length of the scope (1cm comparedto tens of mm in arthroscopes), or there may be performance benefits associated63Table 3.1: Results from the Mann-Whitney U Test, displayed as p-valuesfrom each pairwise comparison. * indicates significance following theBonferroni-Holm correction for multiple comparisons.Bracing Level STS Fulcrum Armrest RCM RCM + SpringProcedure Length Unbraced 0.001* 0.013 0.00043* 0.006 0.004*STS . 0.386 0.507 0.225 0.261Fulcrum . . 0.525 1.000 0.261Armrest . . . 0.326 0.19RCM . . . . 0.16RMS Error Unbraced 0.002* 0.014 0.007 0.057 0.004*STS . 0.133 0.954 0.204 0.075Fulcrum . . 0.149 0.817 0.019Armrest . . . 0.236 0.082RCM . . . . 0.022Figure 3.19: A comparison of typical scope views during the STS (left) andunbraced conditions (right).with a wider field of view. Traditional arthroscopes provide a wider field of viewby radially distorting the scope view by a ?fish-eye? lens, as shown in Figure 3.2.The spring constant from Section 3.2.4 was compared to stiffer and softersprings in a pilot study reported in Appendix D, and was revised to 3000 N/m,based on reductions in both RMS error and procedure length compared with theSTS condition, and the existing spring constant. This will have the effect of mak-ing the tool more stiff as the subject removes the lesion, potentially reducing over-64Figure 3.20: A conceptual comparison of the brace designs presented so farin this study, mapped on a chart comparing motion type on the x-axisand effective stiffness at the hand on the y-axis.resection without increasing procedure length.In order to better understand the relationship between brace design and per-formance, the range of braces studied so far were plotted on a chart comparingmotion type and the conceptual effective stiffness at the hand, Kh(Figure 3.20).The unbraced case sits in the bottom right, where motion is not reversed and nosupport is given to the tool or hand, resulting in no supplemental hand stiffness.Conversely, on the far left are the two types of fulcrum braces presented earlier:Fulcrum1 from Chapter 2 and Fulcrum2 from Section 3.2.1. Both of these fulcrumbraces exhibited motion reversal but Fulcrum2 added support from simulated softtissue.Moving inwards, we see that the STS provides some motion reversal, thoughnot as rigid as the fulcrums, while adding some stiffness at the hand. The armrestallowed users the ability to counteract some of the motion reversal of the foam,likely due to the added stiffness at the wrist. The subjects tended to bias their armagainst the armrest during use, meaning that ad/abduction of the wrist helped toprovide tool movement before the arm was positioned, essentially meaning that the65wrist acted as the instantaneous centre of motion while the forearm was stationary.Had the armrest been able to move with the forearm during use, similar to the han-drest from Fehlberg et al. [2012], we would likely see a different behaviour, suchas motion at the shoulder, relying on being able to displace the armrest, governinglesion resection, being adopted.The far right, upper portion of 3.20 is where the RCM-based braces are lo-cated. The RCM is considered to be more stiff than the Fulcrum conditions becausethe tool must deform more soft tissue to achieve the same tooltip displacement.Further, adding the spring to introduce axial stiffness resulted in more effectivestiffness at the hand.Comparing the performance trends from Figure 3.18 to Figure 3.20, it appearsthat adding stiffness at the hand improves performance compared to an unbracedconfiguration, and that motion reversal can impair performance as suggested byHodgson et al. [1999]. Further studies are needed, perhaps in a less complex task,to explore the impact that effective hand stiffness and motion reversal can have onperformance.In the following study, described in the next chapter, the best performing brace,the RCM+Spring shown in Figure 3.18, was selected to be compared against thesurgical simulation condition (STS only) and a control (Unbraced) in a larger userstudy to obtain statistical power for multiple comparisons of the effects of bracingon procedure length and RMS error on skilled and unskilled subjects.66Chapter 4The Design and Testing of aBracing Strategy for SimulatedFemoral Head-NeckOsteochondroplastyThis chapter was submitted for publication.4.1 AbstractThis study investigated the potential value of bracing in the context of bone millingto treat cam-type femoroacetabular impingement (FAI) lesions. This procedurewas chosen due to the growing recognition of FAI as a mechanism for the devel-opment of early osteoarthritis. The goal of this study was to evaluate whether aproposed bracing technique could enable a user to perform a simulated cam resec-tion faster and more accurately than a currently employed arthroscopic technique.20 non-surgeons and 5 experienced arthroscopic surgeons performed simulatedcam resections under three different conditions: (1) Free Cutting, where the toolreceived no support, (2) Soft Tissue Support (STS), where the tool was supportedby simulated soft tissue, and (3) Braced, where the tool was supported by a me-chanical link that eliminated tool reversal and provided additional axial stiffness.67Subjects performance was assessed for procedure length and accuracy of resection.Bracing significantly reduced both error and procedure length in non-surgeonsubjects when compared to the Free Cutting condition (on the order of 30 and 37%,respectively) but no statistically significant differences were detected compared tothe STS condition, nor between any conditions in the smaller surgeon subject pool.Bracing enabled non-surgeon subjects to perform the task as quickly and accu-rately as experienced arthroscopic surgeons, which suggests that bracing may bebeneficial for inexperienced surgeons performing selected arthroscopic procedureswhich lack significant soft tissue support.4.2 IntroductionFemoroacetabular impingement (FAI) is a deformity of the proximal femur and/oracetabulum which causes pain during articulation, particularly with flexion and in-ternal rotation (Ito et al. [2001]). It has recently been shown to cause early-onsetosteoarthritis (OA) in a prospective cohort study (Agricola et al. [2012]), whichsupports earlier hypotheses to this effect (Leunig et al. [2009] Ganz et al. [2003]Ganz et al. [2008]). There are three primary types of FAI: cam, pincer, and mixed/-combined (a combination of the two former types). Cam impingement is the resultof a non-spherical portion of the femoral head deforming the articular surface ofthe acetabulum, resulting in reduced range of motion, which can eventually lead tofailure of the articular surface of the joint. This deformity can be treated by arthro-scopic surgery in which the bony lesion is resected using a handheld milling toolequipped with a spherical burr. It has been shown that the accuracy of resection ofthe cam lesion affects the surgical outcome(Philippon et al. [2007] Sampson [2001]Willimon et al. [2011]) and that 79-92% (Philippon et al. [2007] Sampson [2005])of revision hip arthroscopies are caused by inadequate resection of the cam lesion(Heyworth et al. [2007]), indicating a need for more accurate resection. Bracingis a strategy that people use either intentionally or subconsciously to improve per-formance in a variety of tool-using tasks, such as robotic grinding Asada and West[1985]. A more common example is when people brace their hand against a sur-face during writing, illustrated in Figure 4.1, which allows more precise motion tobe executed primarily by the degrees of freedom in the hand, rather than the entire68Figure 4.1: Two styles of handwriting. Unsupported (A) and supportedthrough palm contact on the writing surface (B).arm. Bracing can be defined as a parallel mechanical link between the actor, theenvironment, and/or the workpiece that alters the mechanical impedance betweenthe tool and workpiece in order to improve task performance. Such a link createsa secondary, parallel load path which can increase end-effector stability, stiffness,and may reduce fatigue if a person is performing the task instead of a robot manip-ulator.In this study we investigated the potential value of bracing in the context ofthe surgically relevant 3 DoF task of bone milling with a spherical burr. Morespecifically, we simulated the milling of cam-type lesions, as described by Ganzet al. [2003]. We chose this procedure due to the growing recognition of FAI asa mechanism for development of early onset OA (Philippon et al. [2007]) and theconsequent importance of accurate resection to prevent OA. As demand for sur-gical intervention increases (Montgomery et al. [2013]), surgeons are reluctant toembrace technologies that require increased procedure time. We hypothesize thatbracing can enhance precision in femoral neck osteoplasty without extending pro-cedure time. The goal of this study was to evaluate whether a proposed bracingtechnique could enable a user to perform a simulated surgical resection more accu-rately and quickly than a currently employed arthroscopic technique (Byrd [2005]).69Figure 4.2: The simulated cam lesion used in the study. The poplar woodsubstrate is 1.5x1.5. The black representative cam lesion is 19mm indiameter, 3.6mm thick at the centre, with a weight of 0.6 +/- 0.05 g.4.3 Materials and MethodsOur primary interest was investigating the accuracy of milling during the finalphase of bone removal where only a small amount of lesion remained as this waswhere over-resection was most likely to occur. A small domed shape that simulatedthe curved contour of a cam lesion (19mm diameter, 3.6mm thick at the centre, 0.6+/- 0.05 g), shown in Figure 4.2, was created from two-part black urethane plasticusing a silicone mold casting process and glued onto a 1.5x1.5 poplar wood sub-strate. The contrast between the black urethane of the mock cam lesion and thelight colour of the poplar wood provided a clear boundary for subjects to use inidentifying the desired limits of resection and was inspired by the colouring usedin various computer-assisted surgical systems (e.g. MAKO and NavioPFS). TheShore D hardness of the urethane plastic was reported as 75D, providing similarproperties to human cortical bone (75-85D) found by Kincaid et al. [2007]. Testsubjects included 20 adults (11 male, 9 female, 29, SD 8 years) with no surgicalexperience and 5 orthopaedic surgeons (male, 35, SD 7 years) who had performedan average of 115 arthroscopic surgeries [Range: 0 to 275]. Subjects were in-structed to use a USB borescope (simulated arthroscope) and a handheld shaverwith 5.5mm abrader burr (Smith & Nephew Dyonics) to remove the simulated camlesion as quickly as possible with due respect for accuracy, under varying levels oftool support. Three types of tool support were tested, as shown in Figure 4.3: (A)Free Cutting The sample was clamped in a box, covered with black lycra fabric70Figure 4.3: Subjects were tested under three different conditions: (A) FreeCutting - The tool received no support. (B) Soft Tissue Support (STS) -The test box was filled with a low and high density foam to simulate thesupport provided by soft tissue and the joint capsule. (C) Braced TheSTS condition was combined with a mechanical brace that eliminatedthe reversal of motion of the tool that occurs in the STS condition, aswell as provided axial stiffness through a compression spring, makingthe tool stiffer as it approached the sample.which provided no support to the tools, with access ports for the USB borescopeand handheld shaver. (B) Soft Tissue Support (STS) The sample was clampedin a box identical to (A), with the addition of a foam-based soft tissue interface(developed in consultation with our advising surgeon) which simulated the supportprovided by the musculature and joint capsule, including a vertical slit mimick-ing a capsulotomy, which was placed between the tool and the sample. The foamstiffness was chosen to approximate the tissue stiffness we measured in an ex vivoporcine model. (C) Braced - Motion reversal caused by operating across a fulcrum,as is performed in typical minimally invasive surgery, has been shown to impairperformance during laparoscopic tasks (Hodgson et al. [1999]). In order to correctfor this motion reversal, an instantaneous centre of rotation was placed below thetarget area, eliminating the motion reversal by mechanically requiring that the toolrotate around the location of the spherical bearing beneath the workspace. Further,the tool was stiffened axially by adding a 3000 N/m spring to the spherical bear-ing to resist axial translation and thereby reduce any tendency to over-resect thelesion. The zero-point of the spring was set so that the tip of the milling tool hadjust penetrated the simulated joint capsule, and further axial movement involved71Figure 4.4: Rapidform XOR2 was used to process raw scan data into a planar3D model. Scan data was loaded using the mesh buildup feature (A),aligned using a globally referenced iterative closest point algorithm (B),merged into a single model (C), and finalized by redistributing the polyfaces (D). The final model used for comparison is shown with manuallytrimmed edges in (E). Rapidform XOV3 was used to analyze the RMSroughness as calculated by a plane fit to the surface of the sample (F).compression of the spring.Subjects performed 4 resections per bracing condition, in a randomized order toaccount for any effects of learning, for a total of 12 samples per subject. Subjectswere timed by starting a timer when the tooltip contacted the lesion and endingwhen the entire lesion had been removed and the tool withdrawn. Task posturewas constrained to mimic an arthroscopic surgeon using the supine arthroscopyapproach(Dienst [2005]). Upon completion of the study, subjects were asked to fillout a debrief questionnaire, with detailed questions and results shown in Section F.Accuracy of resection was assessed using a VIVID 9i 3D laser scanner (KodakMinolta Sensing Americas, Inc.). Laser scanning was performed with a wide anglelens at a distance of 600mm to the sample. Scans were taken at 30? intervals, rang-ing from +/-60? to ensure that only the front portion of the sample was captured,which reduced the scanning time by approximately 50%, compared with imagingthe front and back of the sample.Post-resection laser scans were transformed into 3D models using Rapidform72XOR2 (3D Systems Corporation). Scan data was loaded into Rapidform XOR2(Figure 4.4 A) using the mesh buildup feature, aligned using a globally referencediterative closest point algorithm (Figure 4.4 B), merged into a single model (Fig-ure 4.4 C), and finalized by redistributing the poly faces (Figure 4.4 D). The finalmodel, shown in detail in Figure 4.4 E, had the edges trimmed manually, to elimi-nate extraneous information during mesh deviation analysis.A reference plane was fit to the planar surface of the wood substrate by pickingseveral points on the surface unaffected by the lesion removal. A mesh-deviationcalculation from this reference plane was performed, illustrated in Figure 4.4 F.The results are reported as RMS deviations.A scan-rescan repeatability analysis for the laser scanning protocol found thatthe upper 95% error bound on the RMS error calculation was 0.037 mm. A re-peatability analysis of the reference plane fitting feature, found that two standarddeviations, representing the variability of 95% of the data, was 0.021mm, which isless than the upper 95% error bound reported in the scan-rescan analysis. As such,differences greater than 0.04mm are likely to be real.The statistical analysis package, IBM SPSS 20, was used to analyze the resultsof the present study. We first tested the null hypothesis that the data came from anormally distributed population using the Shapiro-Wilk test and found significantresults for both procedure length (p < 0.0001) and RMS error (p < 0.0001), in-dicating that non-parametric tests must be used. The per sample time and RMSdeviation data was then rank-transformed, with the smallest value being given rank1. The Kruskal-Wallis test for independent samples detected a significant effect oftrial number on procedure length (p < 0.0001) but not RMS error (p = 0.960) whendata from all 4 trials was included. When data from the first trial was excluded (andremaining data re-ranked), there was no detectable effect of trial number on eitherprocedure length (p = 0.284) and RMS error (p = 0.963). Using data from the final3 trials only, the Kruskal-Wallis test detected a significant effect of bracing on bothprocedure length (p < 0.0001) and RMS error (p < 0.014), so we reject the null hy-pothesis that the rank distribution of our variables was the same across all levels ofbracing. The Mann-Whitney U Test for pairwise comparisons, with a Bonferroni-Holm (Holm [1979]) correction for multiple post-hoc comparisons, was used todetermine which type(s) of bracing affected procedure length and RMS error.73Figure 4.5: Results are shown as the mean +/- SD, compared to the idealsurgical intervention in the bottom left corner.4.4 ResultsThe effects of bracing on time and accuracy of simulated femoral neck osteoplastyare shown in Figure 4.5 with data presented as the mean and standard deviation forboth the 20 non-surgeon subjects and the 5 surgeons.Bracing reduced deviations from the target resection in the non-surgeon groupby 30% for the STS (p < 0.012) and Braced (p < 0.011) conditions compared tothe Free Cutting control, although no significant differences were detected betweenthe two supported conditions. There was a nominal reduction in RMS error for theSTS condition of approximately 30% compared to the Bracing and Free Cuttingconditions, however no statistically significant differences were found.Bracing reduced procedure length in the non-surgeon group by 37% for theSTS (p < 0.0001) and Braced (p < 0.0001) conditions compared to the Free Cut-ting control, although no significant differences were detected between the twosupported conditions. There was an approximately 10% reduction in procedure74Figure 4.6: Results comparing the average deviation as a function of the sup-port condition, shown as the mean +/- SD, compared to the ideal surgicalintervention at the dashed horizontal axis.length for the STS and Braced conditions compared to the Free Cutting control butno statistically significant differences were found.As illustrated in Figure 4.5, adding support to the tool through simulated softtissue or a mechanical brace enabled non-surgeons to perform the task as quicklyas experienced surgeons.The average deviations, shown in Figure 6 as means +/- SD, have a statisticallysignificant bias (p < 0.0001 for both surgeon and non-surgeon groups) of roughly0.1 to 0.2 mm in all conditions. In the non-surgeon population, bracing reducedthis bias by a nominal amount of 30% when compared to the Free Cutting control(p < 0.019), although no statistically significant differences between the STS andFree Cutting or STS and Braced conditions were found. There were no statisticallysignificant differences in average deviation between the support conditions in thesurgeon group.754.5 DiscussionIn this study we investigated the potential value of bracing in the context of bonemilling to treat simulated cam-type FAI lesions because this kind of deformity hasbeen shown to cause early onset OA and developing effective treatments is there-fore of strong current interest. The goal of this study was to evaluate whether aproposed bracing technique could enable a user to perform a simulated cam le-sion resection more accurately and quickly than a currently employed arthroscopictechnique.We demonstrated that bracing can reduce errors associated with over-resectionof a mock cam lesion in a non-surgeon population by a nominal amount of 27%when compared with a control (Free Cutting) condition that provided no support tothe tool or scope, and a similar reduction in procedure time (approximately 37%),but we did not see similar effects in the surgeon group. In all conditions, there wasa small over-resection bias of roughly 0.1-0.2 mm.These findings potentially had significant implications for the treatment of FAI.Between 2004 and 2009, the rate of hip arthroscopy increased by 365% in theUnited States (Montgomery et al. [2013]). As a result of this increased demandfor procedures, surgeons are interested in techniques that improve accuracy anddecrease operating time. Although experienced surgeons did not realize any ben-efits from bracing in this study, tool support did significantly improve both accu-racy and procedure time in the non-surgeon population, which might be reasonablyrepresentative of surgeons-in-training. However, since this group experienced nosignificant supplemental benefit from bracing compared with the soft tissue sup-port condition, bracing in the form we studied is unlikely to provide any significantbenefit for FAI surgery itself. However, based on our results, it is plausible thatbracing may be of value during the learning curve in other arthroscopic procedures(e.g., at the shoulder or knee) where there is less natural soft tissue support for theshaving tool, and this possibility should be investigated further.This study also has implications for the design of computer assisted surgical(CAS) systems designed to support arthroscopic shaving procedures. Such sys-tems have been shown to improve a surgeon?s ability to interoperatively identifyand remove FAI lesions (Ecker et al. [2012]) by providing a display and a tracked76tool that displays a colour-based margin for cam lesion removal. We demonstratedthat there was an over-resection bias of roughly 0.1-0.2 mm for both surgeon andnon-surgeon groups when using a colour change to indicate the material removalboundary, and this bias should be taken into account when designing a CAS guid-ance system.The primary limitations of this study include limits on the fidelity of the sim-ulated surgical task and the limited number of surgeons we were able to recruit.With regard to fidelity, subjects in our simulation were tested on their ability toremove a representative lesion on a flat surface because this facilitated rapid andreproducible production and analysis of the samples. The actual surgical task in-volves a curved surface which would be more difficult to replicate. Also, duringlive surgery, surgeons typically use a 30? arthroscope during the milling phase,whereas we used a USB borescope which has a longer focal length and a smallerfield of view. Changing the viewing condition compared to real surgery may haveimpaired the ability of the subjects to adequately visualize the cam lesion, althoughthe surgeon cohort did not report experiencing any visualization difficulties duringthe study.In responses elicited through a formal debriefing questionnaire (Section F), thesurgeon cohort mentioned that the tools felt as if they had more axial support duringthe procedure than would be typical in surgery. This may have had an impact ontheir handling properties compared with live arthroscopic procedures in which theirrigation and natural fluids would allow the tools to move more freely in and outof the arthroscopic site.Finally, participants frequently reported in the debriefing questionnaire that thebrace felt somewhat too stiff both laterally and axially, despite strong majorities(80% of the non-surgeon group and 75% of the surgeon group) reporting that theyenjoyed using it. We believe this stiffness may have caused a perception of fatiguein the subject?s upper limb, as reported by 55% of the non-surgeon and 30% of thesurgeon group, and could potentially have led to an underestimate of the value ofbracing in this study. We had selected the axial stiffness based on pilot study datainvolving the subject author, who may have not been representative of the sam-ple population, which suggests that the optimal perceived level of axial stiffnessprovided by bracing could be a matter of personal preference.77The small sample size of the surgeon cohort (unfortunately limited by recruit-ment difficulties due to the relatively small number of surgeons performing arthro-scopic procedures) prevented us from achieving sufficient statistical power to drawconclusions both within the surgeon cohort and between the two cohorts. Post-hoc power analysis suggests that a sample size similar to the non-surgeon group isrequired to achieve sufficient power to observe effects between the two groups.Over-resection of bone during cam lesion milling places the patient at risk forfemoral neck fracture(Philippon et al. [2007]). However, a finite element model-ing analysis,validated with cadaveric specimens, has shown that up to 10mm ofthe femoral neck can be removed before it significantly increases the likelihoodof post-operative fractures(Alonso-Rasgado et al. [2012]). This implies that theover-resection bias we found of 0.1-0.2 mm is not likely of concern in a clini-cal setting. In practice, however, over-resection concerns produce a bias towardsunder-resection, which in turn leads to significant revision rates (reported to be inthe range of 1.4% - 4.8% (Clarke et al. [2003] Willimon et al. [2011]), of which 79-92% have been attributed to inadequate primary resection (Philippon et al. [2007]Sampson [2005])).In conclusion, we have found a significant effect of bracing for a non-surgeongroup performing a clinically realistic bone milling task, which suggests that abracing strategy could potentially be of benefit for novice surgeons learning to per-form arthroscopic shaving procedures where significant intrinsic soft tissue supportis not present.4.6 AcknowledgementsWe thank our subjects for their participation in this study, the Centre for Hip Healthand Mobility for the use of their facilities, and the Natural Sciences and Engineer-ing Research Council of Canada for providing funding to support this work.78Chapter 5Discussion and Conclusions5.1 Summary of Thesis ContributionsIn this thesis we applied the concept of bracing to a 3 DOF task in an extension ofprevious work in our lab examining 1 DOF drilling. FAI milling was chosen as areference task because it fit our selection criteria:1. 3 DOF for the tool to provide a logical extension of previous 1 DOF workfrom our lab.2. Suitable surgical complexity to enable abstraction and testing on a phantommodel.3. The ability to be tested using both a surgeon and non-surgeon population.4. Potential applicability to a significant surgical challenge.Further, using FAI as a vehicle for exploring 3 DOF bracing allowed us to collabo-rate with clinical experts and observe the procedure during live surgeries, ensuringthat the study outlined in Chapter 4 would be clinically relevant.Specific contributions of this thesis are:1. A method for creating replicas of human anatomy with different coloureddeformities, as illustrated by the proximal femur segments with cam lesionsfrom Chapter 2.792. One of the first studies explicitly applying bracing to an orthopaedic prob-lem, presented at an international conference (Chapter 2).3. A high fidelity simulation of arthroscopic femoral head-neck osteochon-droplasty, as confirmed by a surgeon cohort, used in Chapters 3 and 4.4. A comprehensive examination of the impact of custom-manufactured braceon a non-surgeon and surgeon population performing a simulated arthro-scopic milling task (4).Our first look at applying bracing to FAI intervention, outlined in Chapter 2,found that bracing could potentially improve task performance but was limited by asmall sample and effect size. The design of the brace, a spherical bearing fulcrum,was potentially unsuitable for the task, as it didn?t provide sufficient workspaceto allow for adequate resection of the cam lesion, which possibly produced anunderestimate of the potential of bracing.We then performed a exploration of the bracing design space as applied to arevised simulation of the surgical procedure, as outlined in Chapter 3. The resultscontinued to provide encouragement that our hypothesis that bracing could im-prove both speed and accuracy of cam lesion resection by untrained subjects maybe true. The end result of the study was the finding that an RCM + Spring brace,detailed in 3.2.4, enabled users to perform the simulated surgery faster and moreaccurately than the surgical simulation condition; however, this hypothesis neededto be tested in a larger population in order to achieve sufficient statistical power.We recruited 20 non-surgeons and 5 experienced arthroscopic surgeons to eval-uate the effects of the RCM + Spring brace in comparison with the surgical simula-tion (Foam) and unbraced conditions. When compared against the unbraced con-dition, our brace reduced procedure length and error by 37% and 27% respectivelyin the non-surgeon population, however there were negligible differences when thebrace was compared to the Foam condition. This finding suggests that an optimallevel of bracing may exist, and diminishing or even negative returns may be real-ized when this level is exceeded. How to determine the optimal level of bracingremains an unanswered question.Interestingly, this finding of benefit from bracing did not carry over to the sur-geon group, where our results suggest that experienced surgeons may not require80additional support to accurately remove the cam lesion when a clear boundary forresection exists, or that the RCM + Spring brace was not an effective brace designfor this task. The brace design, projecting an RCM beneath the target area, provideslateral motion scaling by requiring that the hand move a father distance than thetooltip (which is closest to the RCM), providing an increase in the lateral preci-sion of the tool. This type of precision increase alone, the RCM brace from 3.2.3,provided no nominal performance improvement over the STS condition (shownin Figure 3.18), suggesting that more attention be paid to the axial direction. In-troducing axial stiffness provided nominal improvement over the STS condition,in Chapter 3, by increasing the force required for axial displacement as the toolmoved closer to the target cam lesion. However, these effects did not carry over toa larger population, shown in Chapter 4.Further exploration of the bracing design space should be performed, lookingat variables beyond the effective stiffness at the hand (as shown in Figure 3.20).Of particular interest is the concept of axial motion scaling, rather than axial stiff-ness, to reduce over-resection errors. We believe that over-resection errors, in thisstudy, are due to both an inability to precisely determine the interface between un-der and over-resection during the cutting procedure, when the tooltip obscures thearthroscope, and an inability in an unskilled/less-skilled population to effectivelymodulate arm stiffness to compensate for unanticipated tooltip disturbances (asshown by Figure 4.5, where the non-surgeon population in an unbraced configu-ration had significantly worse performance compared to supported conditions, andin comparison to the surgeons who had somewhat uniform performance across allconditions). If we could provide a subject with the ability to place the tooltip at theprecise interface between over and under-resection, through motion scaling or anintelligent axial impedance that adjusts itself based on the distance from the tooltipto the target, then the error associated with the milling task could be reduced. Ofcourse, this depends on the ability of a CAS system to display the required infor-mation interoperatively.The ability to display a clear margin for tissue removal during arthroscopicsurgery is just beginning to emerge (Nawabi et al. [2013]). One of the first groupsto track arthroscopic tools interoperatively was Monahan and Shimada [2008], whoused an encoder linkage system during a simulated navigation task rather than a81surgical simulation. The linkages were calibrated preoperatively using CAD data,which could be replaced by a CT scan. They showed that using the CAS systemresulted in a 38% reduction in task completion time and 78% reduction in tool pathlength.Brunner et al. [2009] performed a prospective study that randomized partici-pants into navigated and non-navigated arthroscopic cam lesion treatment groups.The study used the alpha angle to determine the amount of resection in the nav-igated group. The alpha angle is formed by a line drawn parallel to the femoralneck axis, from a superior perspective, and a line from the center of the femoralhead to the transition of femoral head into the femoral neck. They were unable toshow any significant effects of navigation on the outcome measures, highlightingthe limitations of the alpha angle as an outcome measure, and the importance ofaccurate preoperative planning.Most recently, a group (Ecker et al. [2012]) using a heavily modified versionof the software from Brunner et al. [2009], showed that the discrepancy betweenplanned and actual reaming was less than 1mm on average in 18 operations, per-formed by two surgeons.The margin for tissue removal in this work was provided by a clear contrastbetween the black cam lesion, and the white urethane plastic or light colouredpoplar wood. Commercially available CAS systems often use a staged colour codeto indicate how much tissue remains before over-resection occurs, whereas oursystem offered no warning prior to over-resection. If an intermediate layer of adifferent colour were introduced in the boundary between over and under-resection,with a thickness that was roughly equivalent to the average over resection errorfrom Figure 4.6, additional resection accuracy could be realized.Future work should examine the effectiveness of computer assistance and pre-operative planning on the ability of surgeons to accurately remove a cam lesion inlive surgery to better understand if there is a potential application for bracing inFAI surgery.825.2 Applying Bracing to Other Orthopaedic ProceduresWhile the tool may receive adequate support from the surrounding soft tissue inFAI interventions, there are a number of other procedures that may benefit fromfuture study, namely arthroscopic procedures on joints that provide less support tothe tool, such as the shoulder or the knee. We briefly discuss these applicationsbelow.5.2.1 Arthroscopic Acromioplasty (Shoulder)In brief, the shoulder can also be subject to impingement syndromes as first de-scribed by Neer II [1972]. However, rather than bony lesions as we?ve presented inthis thesis, shoulder impingement arises from contact between an extension of theshoulder blade called the acromion and the supraspinatus tendon when the arm islifted above shoulder height.While non-operative treatment is preferred (e.g. by strengthening the rotatorcuff and correcting scapula posture), some patients do require arthroscopic surgeryto correct this problem (Sampson et al. [1991]).In such a procedure, the arthroscope is inserted into the subacromial spaceabove the supraspinatus tendon, as illustrated in Figure 5.1, and the subacromialdefect is evaluated. The shaver is inserted through a portal either laterally or pos-teriorly, as shown in Figure 5.31, and the defect is removed with a high speed burrsimilar to the one used in this thesis.Similarly to the rising demand for hip arthroscopy, there is also a growingdemand for acromioplasty. One study reports a 142% growth in demand between1999 and 2008 (Vitale et al. [2010]), which implies that there will be significantinterest in approaches that can reduce procedure length. We therefore believe thatthere is potential for the use of bracing in shoulder procedures.1Reprinted from Arthroscopy, 7(3), TG Sampson et al., Precision Acromioplasty in ArthroscopicSubacromial Decompression of the Shoulder, 301-307, Copyright (1991), with permission from El-sevier83Figure 5.1: The left shoulder and acromioclavicular joint. The musculatureis not shown. Grey [1918]5.2.2 Unicompartmental Knee ArthroplastyThe most common arthroplasty procedure in Canada is the TKA, with over 20 000performed in 2009-2010 [CIHI].In this procedure, the distal end of the femur and the proximal end of the tibiaare exposed and a cavity is created to accept an implant.The TKA has an excellent 15 year survival rate of 94-98% (Meek et al. [2004]),although this survival rate drops to 76% at 10 years if implanted in a patientyounger than 60 (Rand and Ilstrup [1991]). This is thought to occur as a resultof younger patients tending to have a less advanced stage of OA and placing ahigher functional demand on their implants through higher activity levels.84Figure 5.2: The left shoulder with musculature, emphasizing the rotator cuffand the supraspinatus muscle superior to the spine of the scapula. Grey[1918]One study reported that between 5-20% of patients who received a TKA hadarthritis in only one knee compartment (Satku [2003]), and Marmor [1988] arguedthat the removal of healthy tissue is contrary to the basic orthopaedic principle ofpreserving normal structures whenever possible . As such, UKA has been receivingrenewed interest.In a typical UKA procedure, a smaller incision is made on either the medial orlateral side of the knee, depending on the compartment that needs to be replaced.Mechanical jigs are fixed to the tibia and femur, and the diseased bone is removedwith a saw and replaced with an implant.The strongest prognostic indicator for the long-term survival of the UKA im-plant is the accuracy of implantation (Cartier et al. [1996]). However, it has beenshown that conventional cutting jigs produce unsatisfactory accuracy 30% of the85Figure 5.3: Sagittal view of the shoulder, with arthroscopic milling tool enter-ing posteriorly to remove the shaded portion of the acromion. Sampsonet al. [1991]time (Tabor Jr and Tabor [1998]), and that saw blade deflection can play a role inthe accuracy of the cut (Plaskos et al. [2002]). This suggests an important moti-vation for considering shifting towards using bone milling for the bone removalprocess.5.2.3 The MAKOPlasty versus the Precision Freehand SculptorTwo commercially available CAOS options which use a milling process to performUKA are the robotic arm interactive orthopaedic system (RIO) R?(MAKO SurgicalCorp.) and the NavioPFS R?(precision freehand sculptor) (Blue Belt Technologies,INC).The RIO system works on the principle of active constraints, or haptic barriers,imposed through a robotic arm. The patient is given a CT scan which is used toplan the surgery preoperatively, and the robot is registered to the bony anatomy ofthe patient once the surgery begins. The surgeon grasps the milling tool, attachedto the end of the robotic arm shown in Figure 5.4 A, and removes the preoperatively86Figure 5.4: The MAKO RIO (A) and the NavioPFS (B) being used by the the-sis author at the 2013 International Meeting of the International Societyfor Computer Assisted Orthopaedic Surgery.planned area of bone. The robotic arm prevents the surgeon from over-resecting byimposing a virtual wall which limits the workspace.The NavioPFS system, unlike the RIO, does not require preoperative imag-ing, but instead relies on interoperative landmark-based registration and statisticalshape models. The handheld milling tool, similar to what was used in this thesis,is tracked relative to the surgical area and adjusts the burr?s cutting speed to reducethe chance of over-resection. If the surgeon continues to attempt to remove addi-tional bone, the burr will stop cutting and retract inside the tooltip. The device isshown in Figure 5.4 B.These two systems allow different amounts of surgeon control. The RIO re-lies on a very large, rigid robotic arm to prevent excess bone removal, whereas theNavioPFS is more similar to the tools already used in the operating room, in that itis relatively lightweight and able to be held directly by the surgeon. Interestingly,as illustrated in Figure 5.4, while using the NavioPFS, the surgeon is encouragedto establish contact with the free hand between the tool tip and the patient?s kneebone. As defined by this thesis, this is an example of bracing behaviour whichimproves the precision of the tooltip. While a comparison between the accuracy ofthe RIO and NavioPFS has not been made, we speculate that, since both systemsare FDA approved, they have demonstrated similar levels of accuracy. However,the NavioPFS is roughly one third the price of the MAKO system. One study of87interest could examine the impact of bracing a standard milling tool integrated witha tracked surgical display, in comparison with the NavioPFS and the RIO. We sug-gest that bracing alone, even without the additional tooltip retraction mechanismof the NavioPFS or the haptic guidance of the RIO, may enable surgeons to achieveacceptable resection accuracy and so may result in a more cost effective system.We therefore recommend evaluating the potential of bracing in this application.Of course, the preparation of the bony surface for implantation is one com-ponent of a UKA procedure. Surgical outcomes of CAS procedures may dependon registration errors with the tracking system, cutting errors (over/under resec-tion) during implant bed preparation, and errors associated with implant orienta-tion shifting during the cementing phase. Outcome measures for studies examin-ing UKA procedures frequently report deviation from the preoperative plan as themeans assess the accuracy of implantation, and this measure is often the sum of thecontributing errors (tracking, bone cutting, implantation). Bracing in its currentform would predominantly address cutting errors associated with the preparationof the implant bed, so addition performance gains may be realized through combi-nation with navigation technology. Several studies have looked at the use of CASsystems for UKA with their results reported as follows.An analysis of the NavioPFS was performed by Smith et al. [2013], where theyperformed a UKA procedure on twenty (10 left and 10 right) synthetic bone pairs(femur and tibia), without cementing the implants. They first assessed the errorassociated with the optical tracking system, reporting that the RMS deviation was< 0.5mm in all directions. This is comparable to the accuracy limits of the opticaltracking system reporting a resolution of 0.5mm and an RMS error of approxi-mately 0.35mm. They reported that the maximum rotational error was 3.2? andRMS angular error was 1.5? across all orientations, for both the fibia and femoralcomponents. Further, the maximum translational error was 1.2mm and the RMStranslational error was 0.61mm. These results are affected by the cutting error(over and under resection, presumably), and the registration error from the statisti-cal shape modelling approach, however, they did not cement the implants in placewhich limits the ability to compare these results to a study performed in vivo.An analysis of the MAKO RIO was performed by Dunbar et al. [2012], wherethey retrospectively examined a cohort of 20 UKA procedures on patients that met88their inclusion criteria. They reported that surgical RMS errors were within 1.6mmand 3.0? in all directions. The accuracy of the CT-registration system used byMAKO is unknown, but we can assume that the encoders within the robotic armare more accurate than optical tracking systems, suggesting that the majority of theerror associated with this procedure rises from over/under resection and implantshifting during cementing.A study from Cobb et al. [2006], examining the effects of a third CAS sys-tem (ACROBOT) also reported the accuracy of a manual UKA procedure for ref-erence, as they performed UKA procedures on 28 knees randomized to either thecontrol (manual) or ACROBOT system groups. The manual procedure, using non-navigated instrumentation had an average implant translational error in all direc-tions of 2.2mm. This procedure would be predominantly influenced by the abilityof the surgeon to align the traditional instrumentation, in addition to accurate boneremoval.A study by Jenny and Boeri [2003] directly compared the effects of navigationon implant orientation by randomizing 60 patients into two groups: navigated andconventional. On average, all outcome measures were within 2? of the preoper-ative plan and while there was no statistically significant differences between themeasured angles of the two groups(with the exception of the sagittal orientationof the tibial component), the navigated implants tended to be sub-degree in error.Using acceptance criteria of +/-2? deviation from the preoperative plan, navigationhad a statistically significant effect on the success rate of implantation, with 18/30in the navigated group and 6/30 in the conventional group considered acceptable.Our results from Chapter 4 showed that bracing reduced cutting error in a non-surgeon population by 30%. However unlike the studies from Smith et al. [2013],Dunbar et al. [2012], Cobb et al. [2006], and Jenny and Boeri [2003], we were test-ing the subject?s ability to mill a planar surface, which we can expect to be easierthan milling a curved surface for a bone-sparing femoral component. Further com-plicating comparisons between these studies is the inconsistent methodology, aswe?ve discussed previously that implant shifting during cementing is a contributorto the accuracy of implantation. While Smith et al. [2013] reported tremendous ac-curacy from the NavioPFS in a sawbones model, it remains to be demonstrated thatthese effects would hold up in a cadaveric or patient test where surgical conditions89and implantation accuracy would contribute to the outcome of the procedure.Literature shows that CAS technology positively impacts the accuracy of UKAimplantation, be it through robotics, navigation, or a combination of the two. Theimpact of the individual CAS components on the surgical outcome, and the extentto which one has the larger individual contribution to error reduction (navigationor robotics) remains to be determined. Future work might assess the impact ofnavigation on bracing, as compared to navigation-only and braced-only conditions,to determine if bracing may be a suitable technology for a relatively low-cost (ascompared to the MAKO RIO and NavioPFS) UKA CAS system.5.3 Closing ThoughtsThis thesis represents some of the first work in the field of braced orthopaedicsurgery, and demonstrates that bracing has the potential to reduce both bone re-moval error and procedure length, without relying on expensive robotic solutions.In a time where CAOS adoption rates are notoriously low, we feel that brac-ing provides a natural and intuitive approach to augmenting surgeon performancewhich could potentially improve the market traction of the technology.Future work in this area should aim to develop a bracing solution that is suitablefor either arthroscopic acromioplasty or UKA, in order to assess the potential benefitof bracing during live surgery.90BibliographyR. Agricola, M. P. Heijboer, S. M. a. Bierma-Zeinstra, J. a. N. Verhaar,H. Weinans, and J. H. Waarsing. Cam impingement causes osteoarthritis of thehip: a nationwide prospective cohort study (CHECK). Annals of the rheumaticdiseases, pages 1?6, June 2012. ISSN 1468-2060.doi:10.1136/annrheumdis-2012-201643. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22730371. ? pages 13, 14, 68T. Alonso-Rasgado, D. Jimenez-Cruz, C. G. Bailey, P. Mandal, and T. Board.Changes in the stress in the femoral head neck junction afterosteochondroplasty for hip impingement: A finite element study. Journal oforthopaedic research : official publication of the Orthopaedic ResearchSociety, pages 1?8, July 2012. ISSN 1554-527X. doi:10.1002/jor.22164. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22707347. ? pages 23, 78N. Arden and M. C. Nevitt. Osteoarthritis: epidemiology. Best practice &research. Clinical rheumatology, 20(1):3?25, Feb. 2006. ISSN 1521-6942.doi:10.1016/j.berh.2005.09.007. URLhttp://www.ncbi.nlm.nih.gov/pubmed/16483904. ? pages 14H. Asada and Y. Sawada. Design of an adaptable tool guide for grinding robots.Robotics and computer-integrated manufacturing, 2(1):49?54, 1985. URLhttp://www.sciencedirect.com/science/article/pii/0736584585900079. ? pages3, 11H. Asada and H. West. Design and Analysis of Braced Manipulators for improvedstiffness. In Proc. of the 3rd International Symp. on Robotics Research, pages62?67, 1985. URLhttp://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Design+and+Analysis+of+Braced+Manipulators+for+Improved+Stiffness#0.? pages 3, 11, 6891N. Awan and P. Murray. Role of hip arthroscopy in the diagnosis and treatment ofhip joint pathology. Arthroscopy : the journal of arthroscopic & relatedsurgery : official publication of the Arthroscopy Association of North Americaand the International Arthroscopy Association, 22(2):215?8, Feb. 2006. ISSN1526-3231. doi:10.1016/j.arthro.2005.12.005. URLhttp://www.ncbi.nlm.nih.gov/pubmed/16458808. ? pages 20C. Bombardier, G. Hawker, and D. Mosher. THE IMPACT OF ARTHRITIS INCANADA : TODAY AND OVER THE NEXT 30 YEARS. Technical report,Arthritis Alliance of Canada, 2011. ? pages 12, 14, 15W. Book and J. Wang. Technology and Task Parameters Relating to theEffectiveness of the Bracing Strategy. In Proceedings of the NASA Conferenceon Space Telerobotics, 1989. URLhttp://smartech.gatech.edu/handle/1853/39035. ? pages 4, 11W. Book, V. Sangveraphunsiri, and S. Le. The bracing strategy for robotoperation. In 5th Joint IFToMM-CISM Symposium on the Theory of Robots andManipulators (RoManSy), volume 84, 1984. URLhttp://smartech.gatech.edu/handle/1853/39191. ? pages 3, 11, 28W. J. Book. New Concepts in Lightweight Arms. In Second InternationalSymposium on Robotics Research, pages 203?205, Tokyo, Japan, 1984.Georgia Institute of Technology MIT Press. URLhttp://hdl.handle.net/1853/39165. ? pages 3, 11I. B. Botser, T. W. Smith, R. Nasser, and B. G. Domb. Open surgical dislocationversus arthroscopy for femoroacetabular impingement: a comparison of clinicaloutcomes. Arthroscopy : the journal of arthroscopic & related surgery : officialpublication of the Arthroscopy Association of North America and theInternational Arthroscopy Association, 27(2):270?8, Feb. 2011. ISSN1526-3231. doi:10.1016/j.arthro.2010.11.008. URLhttp://www.ncbi.nlm.nih.gov/pubmed/21266277. ? pages 19, 20M. a. Bredella, E. J. Ulbrich, D. W. Stoller, and S. E. Anderson. Femoroacetabularimpingement. Magnetic resonance imaging clinics of North America, 21(1):45?64, Feb. 2013. ISSN 1557-9786. doi:10.1016/j.mric.2012.08.012. URLhttp://www.ncbi.nlm.nih.gov/pubmed/23168182. ? pages xi, 19G. Brisson, T. Kanade, A. M. DiGioia III, and B. Jaramaz. Precision FreehandSculpting of Bone. In 7th International Conference Medical Image ComputingComputer-Assisted Intervention, volume 11, pages 105?112. Springer Berlin /92Heidelberg, 2004. URL http://www.ri.cmu.edu/pubs/pub 4982.html. ? pages25A. Brunner, M. Horisberger, and R. F. Herzog. Evaluation of a computedtomography-based navigation system prototype for hip arthroscopy in thetreatment of femoroacetabular cam impingement. Arthroscopy : the journal ofarthroscopic & related surgery : official publication of the ArthroscopyAssociation of North America and the International Arthroscopy Association,25(4):382?91, Apr. 2009. ISSN 1526-3231. doi:10.1016/j.arthro.2008.11.012.URL http://www.ncbi.nlm.nih.gov/pubmed/19341925. ? pages 24, 82J. W. T. Byrd. Hip Arthroscopy, The Supine Approach: Technique and Anatomyof the Intraarticular and Peripheral Compartments. Techniques inOrthopaedics, 20(1):17?31, Mar. 2005. ISSN 0885-9698.doi:10.1097/01.bto.0000152172.34187.bf. URLhttp://content.wkhealth.com/linkback/openurl?sid=WKPTLP:landingpage&an=00013611-200503000-00004. ? pages 29, 69J. W. T. Byrd and K. S. Jones. Arthroscopic femoroplasty in the management ofcam-type femoroacetabular impingement. Clinical orthopaedics and relatedresearch, 467(3):739?46, Mar. 2009. ISSN 1528-1132.doi:10.1007/s11999-008-0659-8. URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2635454&tool=pmcentrez&rendertype=abstract. ?pages xi, 18P. Cartier, J. L. Sanouiller, and R. P. Grelsamer. Unicompartmental kneearthroplasty surgery. 10-year minimum follow-up period. The Journal ofarthroplasty, 11(7):782?8, Oct. 1996. ISSN 0883-5403. URLhttp://www.ncbi.nlm.nih.gov/pubmed/8934317. ? pages 85S. K. Chauhan, R. G. Scott, W. Breidahl, and R. J. Beaver. Computer-assistedknee arthroplasty versus a conventional jig-based technique. The Journal ofBone and Joint Surgery, 86(3):372?377, Apr. 2004. ISSN 0301620X.doi:10.1302/0301-620X.86B3.14643. URLhttp://www.bjj.boneandjoint.org.uk/cgi/doi/10.1302/0301-620X.86B3.14643. ?pages 25(CIHI). Hip and Knee Replacements in Canada - Canadian Joint ReplacementRegistry (CJRR) 2008-2009 Annual Report. Technical report, CanadianInstitute for Health Information, Ottawa, 2009. ? pages 84M. T. Clarke, a. Arora, and R. N. Villar. Hip arthroscopy: complications in 1054cases. Clinical orthopaedics and related research, (406):84?8, Jan. 2003. ISSN930009-921X. doi:10.1097/01.blo.0000043048.84315.af. URLhttp://www.ncbi.nlm.nih.gov/pubmed/12579004. ? pages 78J. C. Clohisy, G. Baca, P. E. Beaule?, Y.-J. Kim, C. M. Larson, M. B. Millis, D. a.Podeszwa, P. L. Schoenecker, R. J. Sierra, E. L. Sink, D. J. Sucato, R. T.Trousdale, and I. Zaltz. Descriptive Epidemiology of FemoroacetabularImpingement: A North American Cohort of Patients Undergoing Surgery. TheAmerican journal of sports medicine, May 2013. ISSN 1552-3365.doi:10.1177/0363546513488861. URLhttp://www.ncbi.nlm.nih.gov/pubmed/23669751. ? pages 16J. Cobb, J. Henckel, P. Gomes, S. Harris, M. Jakopec, F. Rodriguez, a. Barrett,and B. Davies. Hands-on robotic unicompartmental knee replacement: aprospective, randomised controlled study of the acrobot system. The Journal ofbone and joint surgery. British volume, 88(2):188?97, Feb. 2006. ISSN0301-620X. doi:10.1302/0301-620X.88B2.17220. URLhttp://www.ncbi.nlm.nih.gov/pubmed/16434522. ? pages 89A. P. Cooper, S. Z. Basheer, R. Maheshwari, L. Regan, and S. S. Madan.Outcomes of hip arthroscopy. A prospective analysis and comparison betweenpatients under 25 and over 25 years of age. British journal of sports medicine,July 2012. ISSN 1473-0480. doi:10.1136/bjsports-2012-091028. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22797420. ? pages 19M. E. Cosaboom-FitzSimons, S. L. Tolle, M. L. Darby, and M. L. Walker. Effectsof 5 different finger rest positions on arm muscle activity during scaling bydental hygiene students. Journal of dental hygiene : JDH / American DentalHygienists? Association, 82(4):34, Jan. 2008. ISSN 1553-0205. URLhttp://www.ncbi.nlm.nih.gov/pubmed/18755067. ? pages 11A. J. Cossey and A. J. Spriggins. The use of computer-assisted surgical navigationto prevent malalignment in unicompartmental knee arthroplasty. The Journal ofarthroplasty, 20(1):29?34, Jan. 2005. ISSN 0883-5403.doi:10.1016/j.arth.2004.10.012. URLhttp://www.ncbi.nlm.nih.gov/pubmed/15660057. ? pagesM. Dienst. Hip arthroscopy: Technique and anatomy. Operative Techniques inSports Medicine, 13(1):13?23, Jan. 2005. ISSN 10601872.doi:10.1053/j.otsm.2004.09.009. URLhttp://linkinghub.elsevier.com/retrieve/pii/S1060187204000681. ? pages xi,20, 21, 22, 7294S. Dimmick, M. Hons, K. J. Stevens, D. Brazier, and S. E. Anderson.Femoroacetabular Impingement. Radiologic Clinics of North America, 51(3):337?352, 2013. ISSN 0033-8389. doi:10.1016/j.rcl.2012.12.002. URLhttp://dx.doi.org/10.1016/j.rcl.2012.12.002. ? pages xii, 24H. Dong, A. Barr, P. Loomer, and D. Rempel. The effects of finger rest positionson hand muscle load and pinch force in simulated dental hygiene work. Journalof dental education, 69(4):453?60, Apr. 2005. ISSN 0022-0337. URLhttp://www.ncbi.nlm.nih.gov/pubmed/15800259. ? pages 9, 11N. J. Dunbar, M. W. Roche, B. H. Park, S. H. Branch, M. a. Conditt, and S. a.Banks. Accuracy of dynamic tactile-guided unicompartmental kneearthroplasty. The Journal of arthroplasty, 27(5):803?8.e1, May 2012. ISSN1532-8406. doi:10.1016/j.arth.2011.09.021. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22088782. ? pages 88, 89T. M. Ecker, M. Puls, S. D. Steppacher, J. D. Bastian, M. J. B. Keel, K. a.Siebenrock, and M. Tannast. Computer-assisted femoral head-neckosteochondroplasty using a surgical milling device an in vitro accuracy study.The Journal of arthroplasty, 27(2):310?6, Feb. 2012. ISSN 1532-8406.doi:10.1016/j.arth.2011.03.048. URLhttp://www.ncbi.nlm.nih.gov/pubmed/21621956. ? pages 24, 76, 82L. Ejnisman. Femoroacetabular impingement: the femoral side. Clinics in sportsmedicine, 30:369?377, 2011. doi:10.1016/j.csm.2010.12.007. URLhttp://www.ncbi.nlm.nih.gov/pubmed/21419961. ? pages xii, 23R. Elmslie. Remarks on aetiological factors in osteo-arthritis of the hip-joint.British Medical Journal, 1(3757):5?8, 1933. URLhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC2368170/. ? pages 15M. a. Fehlberg, B. T. Gleeson, L. C. Leishman, and W. R. Provancher. ActiveHandrest for precision manipulation and ergonomic support. In 2010 IEEEHaptics Symposium, pages 489?496. Ieee, Mar. 2010. ISBN978-1-4244-6821-8. doi:10.1109/HAPTIC.2010.5444613. URLhttp://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=5444613. ?pages xi, 9, 10, 11M. a. Fehlberg, B. T. Gleeson, and W. R. Provancher. Active Handrest: A largeworkspace tool for precision manipulation. The International Journal ofRobotics Research, 31(3):289?301, Mar. 2012. ISSN 0278-3649.doi:10.1177/0278364911432895. URLhttp://ijr.sagepub.com/cgi/doi/10.1177/0278364911432895. ? pages 9, 11, 6695T. K. Fehring, S. M. Odum, J. L. Troyer, R. Iorio, S. M. Kurtz, and E. C. Lau.Joint replacement access in 2016: a supply side crisis. The Journal ofarthroplasty, 25(8):1175?81, Dec. 2010. ISSN 1532-8406.doi:10.1016/j.arth.2010.07.025. URLhttp://www.ncbi.nlm.nih.gov/pubmed/20870384. ? pages 12, 14R. Ganz, J. Parvizi, M. Beck, M. Leunig, H. No?tzli, and K. a. Siebenrock.Femoroacetabular impingement: a cause for osteoarthritis of the hip. Clinicalorthopaedics and related research, (417):112?20, Dec. 2003. ISSN0009-921X. doi:10.1097/01.blo.0000096804.78689.c2. URLhttp://www.ncbi.nlm.nih.gov/pubmed/14646708. ? pages 13, 14, 16, 19, 28,68, 69R. Ganz, M. Leunig, K. Leunig-Ganz, and W. H. Harris. The etiology ofosteoarthritis of the hip: an integrated mechanical concept. Clinicalorthopaedics and related research, 466(2):264?72, Feb. 2008. ISSN0009-921X. doi:10.1007/s11999-007-0060-z. URLhttp://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2505145&tool=pmcentrez&rendertype=abstract. ? pages 13, 17, 68J. M. Glick and T. G. Sampson. Hip Arthroscopy by the Lateral Approach.Operative Techniques in Orthopaedics, 15(3):218?224, July 2005. ISSN10486666. doi:10.1053/j.oto.2005.07.006. URLhttp://linkinghub.elsevier.com/retrieve/pii/S1048666605000406. ? pages 20H. Grey. Anatomy of the Human Body. Lea & Febiger, Philidelphia, 20 edition,1918. ? pages xi, xvii, 14, 15, 16, 17, 84, 85H. Haider, O. A. Barrera, and K. L. Garvin. Minimally invasive total kneearthroplasty surgery through navigated freehand bone cutting: winner of the2005 ?HAP? PAUL AWARD. The Journal of arthroplasty, 22(4):535?42, June2007. ISSN 0883-5403. doi:10.1016/j.arth.2007.01.010. URLhttp://www.ncbi.nlm.nih.gov/pubmed/17562410. ? pages 25B. E. Heyworth, M. K. Shindle, J. E. Voos, J. R. Rudzki, and B. T. Kelly.Radiologic and intraoperative findings in revision hip arthroscopy. Arthroscopy: the journal of arthroscopic & related surgery : official publication of theArthroscopy Association of North America and the International ArthroscopyAssociation, 23(12):1295?302, Dec. 2007. ISSN 1526-3231.doi:10.1016/j.arthro.2007.09.015. URLhttp://www.ncbi.nlm.nih.gov/pubmed/18063173. ? pages 13, 6896A. J. Hodgson, J. G. Person, S. E. Salcudean, and A. G. Nagy. The effects ofphysical constraints in laparoscopic surgery. Medical image analysis, 3(3):275?83, Sept. 1999. ISSN 1361-8415. URLhttp://www.ncbi.nlm.nih.gov/pubmed/10710296. ? pages 43, 44, 52, 66, 71R. Hollis and R. Hammer. Real and virtual coarse-fine robot bracing strategies forprecision assembly. In Robotics and Automation, 1992. Proceedings., 1992IEEE International Conference on, pages 767?774. IEEE, 1992. URLhttp://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=220276. ? pages x, 4, 5,6, 11S. Holm. A simple sequentially rejective multiple test procedure. Scandinavianjournal of statistics, 6(2):65?70, 1979. URLhttp://www.jstor.org/stable/10.2307/4615733. ? pages 73K. Ito, M. Leunig, S. Werlen, and R. Ganz. Femoroacetabular impingement andthe cam-effect: A MRI-based quantitative anatomical study of the femoralhead-neck offset. Journal of Bone & Joint Surgery, British Volume, 83-B(2):171?176, 2001. ? pages 68J.-Y. Jenny and C. Boeri. Unicompartmental knee prosthesis implantation with anon-image-based navigation system: rationale, technique, case-controlcomparative study with a conventional instrumented implantation. Kneesurgery, sports traumatology, arthroscopy : official journal of the ESSKA, 11(1):40?5, Jan. 2003. ISSN 0942-2056. doi:10.1007/s00167-002-0333-8. URLhttp://www.ncbi.nlm.nih.gov/pubmed/12548450. ? pages 89M. L. H. Jones, R. L. Kirshweng, T. J. Armstrong, and M. P. Reed.Force-Exertion Postures with External Bracing in Industrial Tasks: Data froman Automotive Assembly Plant. In Proceedings of the Human Factors andErgonomics Society Annual Meeting, volume 52, pages 1049?1053, Sept. 2008.doi:10.1177/154193120805201511. URLhttp://pro.sagepub.com/lookup/doi/10.1177/154193120805201511. ? pages9, 12M. L. H. Jones, M. P. Reed, and D. B. Chaffin. The Effect of Bracing Availabilityon Force-Exertion Capability in One-Hand Isometric Pulling Tasks. InProceedings of the Human Factors and Ergonomics Society Annual Meeting,volume 54, pages 1169?1173, Sept. 2010. doi:10.1177/154193121005401517.URL http://pro.sagepub.com/lookup/doi/10.1177/154193121005401517. ?pages 9, 1297M. L. H. Jones, M. P. Reed, and D. B. Chaffin. The effect of bracing availabilityon one-hand isometric force exertion capability. Ergonomics, 56(4):667?81,Jan. 2013. ISSN 1366-5847. doi:10.1080/00140139.2013.765601. URLhttp://www.ncbi.nlm.nih.gov/pubmed/23514040. ? pages xi, 9, 10, 12K. a. Jung, C. Restrepo, M. Hellman, H. AbdelSalam, W. Morrison, and J. Parvizi.The prevalence of cam-type femoroacetabular deformity in asymptomaticadults. The Journal of bone and joint surgery. British volume, 93(10):1303?7,Oct. 2011. ISSN 0301-620X. doi:10.1302/0301-620X.93B10.26433. URLhttp://www.ncbi.nlm.nih.gov/pubmed/21969426. ? pages 16D. Kendoff, M. Citak, V. Stueber, L. Nelson, A. D. Pearle, and F. Boettner.Feasibility of a navigated registration technique in FAI surgery. Archives oforthopaedic and trauma surgery, 131(2):167?72, Feb. 2011. ISSN 1434-3916.doi:10.1007/s00402-010-1114-3. URLhttp://www.ncbi.nlm.nih.gov/pubmed/20490523. ? pages 19D. O. Kendoff, A. Moreau-Gaudry, C. Plaskos, C. Granchi, T. P. Sculco, and A. D.Pearle. A navigated 8-in-1 femoral cutting guide for total knee arthroplastytechnical development and cadaveric evaluation. The Journal of arthroplasty,25(1):138?45, Jan. 2010. ISSN 1532-8406. doi:10.1016/j.arth.2008.11.006.URL http://www.ncbi.nlm.nih.gov/pubmed/19106033. ? pages 25V. Khanduja and R. N. Villar. Arthroscopic surgery of the hip: current conceptsand recent advances. The Journal of bone and joint surgery. British volume, 88(12):1557?66, Dec. 2006. ISSN 0301-620X.doi:10.1302/0301-620X.88B12.18584. URLhttp://www.ncbi.nlm.nih.gov/pubmed/17159164. ? pages 20B. Kincaid, L. Schroder, and J. Mason. Measurement of Orthopedic CorticalBone Screw Insertion Performance in Cadaver Bone and Model Materials.Experimental Mechanics, 47(5):595?607, June 2007. ISSN 0014-4851.doi:10.1007/s11340-007-9056-6. URLhttp://link.springer.com/10.1007/s11340-007-9056-6. ? pages 30, 31, 50, 70B. R. Kivlan, R. L. Martin, and J. K. Sekiya. Response to diagnostic injection inpatients with femoroacetabular impingement, labral tears, chondral lesions, andextra-articular pathology. Arthroscopy : the journal of arthroscopic & relatedsurgery : official publication of the Arthroscopy Association of North Americaand the International Arthroscopy Association, 27(5):619?27, May 2011. ISSN1526-3231. doi:10.1016/j.arthro.2010.12.009. URLhttp://www.ncbi.nlm.nih.gov/pubmed/21663719. ? pages 1998J. Kooyman and A. Hodgson. Tool bracing for performance improvement insimulated femoral head-neck osteochondroplasty. Bone Joint J, 95-B(11):SUPP 28, 2013a. ? pages iv, 28J. Kooyman and A. Hodgson. Tool bracing for performance improvement insimulated femoral head-neck osteochondroplasty. In International Society forComputer Assisted Orthopaedic Surgery (CAOS), page 7, Orlando, 2013b. ?pages iv, 28S. M. Kurtz, E. Lau, K. Ong, K. Zhao, M. Kelly, and K. J. Bozic. Future youngpatient demand for primary and revision joint replacement: national projectionsfrom 2010 to 2030. Clinical orthopaedics and related research, 467(10):2606?12, Oct. 2009. ISSN 1528-1132. doi:10.1007/s11999-009-0834-6. URLhttp://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2745453&tool=pmcentrez&rendertype=abstract. ? pages 12D. Kwon and W. Book. A Framework for Analysis of a Bracing Manipulator withStaged Positioning. In ASME Winter Annual Meeting, Nov. 27-Dec. 2. GeorgiaInstitute of Technology, 1988. URLhttp://smartech.gatech.edu/handle/1853/39040. ? pages 4, 6, 12M. Lavinge, L. Jean-Michel, and V. Pascal-Andre. Historical Evolution of theConcept of Femoroacetabular Impingement as a Cause of Hip Osteoarthritis. InO. Mar??n-Pen?a, editor, Femoroacetabular Impingement, chapter 1, pages 2?6.Springer Heidelberg Dordrecht, London, New York, 1 edition, 2010.doi:10.1007/978-3-642-22769-1. ? pages 16M. Leunig, P. E. Beaule?, and R. Ganz. The concept of femoroacetabularimpingement: current status and future perspectives. Clinical orthopaedics andrelated research, 467(3):616?22, Mar. 2009. ISSN 1528-1132.doi:10.1007/s11999-008-0646-0. URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2635437&tool=pmcentrez&rendertype=abstract. ?pages 13, 68M. Leunig, N. H. Mast, F. M. Impellizerri, R. Ganz, and C. Panaro. Arthroscopicappearance and treatment of impingement cysts at femoral head-neck junction.Arthroscopy : the journal of arthroscopic & related surgery : officialpublication of the Arthroscopy Association of North America and theInternational Arthroscopy Association, 28(1):66?73, Jan. 2012. ISSN1526-3231. doi:10.1016/j.arthro.2011.07.010. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22014697. ? pages 2199J. Lew and W. Book. Hybrid control of flexible manipulators with multiplecontact. Robotics and Automation, 1993., pages 242?247, 1993. URLhttp://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=292153. ? pages 12J. Lew and W. Book. Bracing micro/macro manipulators control. In Robotics andAutomation, 1994. Proceedings., 1994 IEEE International Conference on,pages 2362?2368. IEEE, 1994. URLhttp://ieeexplore.ieee.org/xpls/abs all.jsp?arnumber=350933. ? pages x, 6R. Lung, J. O?Brien, J. Grebenyuk, B. B. Forster, M. De Vera, J. Kopec,C. Ratzlaff, D. Garbuz, H. Prlic, and J. M. Esdaile. The prevalence ofradiographic femoroacetabular impingement in younger individuals undergoingtotal hip replacement for osteoarthritis. Clinical rheumatology, 31(8):1239?42,Aug. 2012. ISSN 1434-9949. doi:10.1007/s10067-012-1981-9. URLhttp://www.ncbi.nlm.nih.gov/pubmed/22552857. ? pages 16, 17L. Marmor. Unicompartmental knee arthroplasty: ten-to 13-year follow-up study.Clinical Orthopaedics and Related Research, 1988. URLhttp://journals.lww.com/corr/abstract/1988/01000/unicompartmental knee arthroplasty ten to.4.aspx. ? pages 85J. B. Mason, J. C. McCarthy, J. O?Donnell, W. Barsoum, M. B. Mayor, B. D.Busconi, V. E. Krebs, and B. D. Owens. Hip arthroscopy: surgical approach,positioning, and distraction. Clinical orthopaedics and related research, (406):29?37, Jan. 2003. ISSN 0009-921X.doi:10.1097/01.blo.0000043041.84315.cc. URLhttp://www.ncbi.nlm.nih.gov/pubmed/12578997. ? pages 20R. M. D. Meek, B. a. Masri, and C. P. Duncan. Minimally invasiveunicompartmental knee replacement: rationale and correct indications. TheOrthopedic clinics of North America, 35(2):191?200, Apr. 2004. ISSN0030-5898. doi:10.1016/S0030-5898(03)00115-9. URLhttp://www.ncbi.nlm.nih.gov/pubmed/15062705. ? pages 84E. Monahan and K. Shimada. Verifying the effectiveness of a computer-aidednavigation system for arthroscopic hip surgery. Studies in health technologyand informatics, 132:302?7, Jan. 2008. ISSN 0926-9630. URLhttp://www.ncbi.nlm.nih.gov/pubmed/18391309. ? pages 23, 81S. R. Montgomery, S. S. Ngo, T. Hobson, S. Nguyen, R. Alluri, J. C. Wang, andS. L. Hame. Trends and demographics in hip arthroscopy in the United States.Arthroscopy : the journal of arthroscopic & related surgery : official100publication of the Arthroscopy Association of North America and theInternational Arthroscopy Association, 29(4):661?5, Apr. 2013. ISSN1526-3231. doi:10.1016/j.arthro.2012.11.005. URLhttp://www.ncbi.nlm.nih.gov/pubmed/23375668. ? pages 12, 69, 76R. O. Murray. The aetiology of primary osteoarthritis of the hip. The Britishjournal of radiology, 38(455):810?24, Nov. 1965. ISSN 0007-1285. URLhttp://www.ncbi.nlm.nih.gov/pubmed/5842578. ? pages 15D. H. Nawabi, D. Nam, C. Park, and A. S. Ranawat. Hip Arthroscopy: The Use ofComputer Assistance. HSS Journal , 9(1):70?78, Jan. 2013. ISSN 1556-3316.doi:10.1007/s11420-012-9313-9. URLhttp://link.springer.com/10.1007/s11420-012-9313-9. ? pages 23, 81C. Neer II. Anterior acromioplasty for the chronic impingement syndrome in theshoulder. The Journal of Bone & Joint Surgery, 54-A(1):41?50, 1972. URLhttp://jbjs.org/article.aspx?articleid=27066. ? pages 83A. Nilsdotter and A. Bremander. Measures of hip function and symptoms: HarrisHip Score (HHS), Hip Disability and Osteoarthritis Outcome Score (HOOS),Oxford Hip Score (OHS), Lequesne Index of Severity for Osteoarthritis of theHip (LISOH), and American Academy of Orthopedic Surgeons (A. Arthritiscare & research, 63 Suppl 1(November):S200?7, Nov. 2011. ISSN 2151-4658.doi:10.1002/acr.20549. URL http://www.ncbi.nlm.nih.gov/pubmed/22588745.? pages 19R. Papalia, A. Del Buono, F. Franceschi, A. Marinozzi, N. Maffulli, andV. Denaro. Femoroacetabular impingement syndrome management:arthroscopy or open surgery? International orthopaedics, 36(5):903?14, May2012. ISSN 1432-5195. doi:10.1007/s00264-011-1443-z. URLhttp://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3337119&tool=pmcentrez&rendertype=abstract. ? pages 19, 20A. Papavasiliou and N. Bardakos. Complications of arthroscopic surgery of thehip. Bone and Joint Research, 1(7):131?144, 2012. URLhttp://www.bjr.boneandjoint.org.uk/content/1/7/131.short. ? pages 13M. J. Philippon, A. J. Stubbs, M. L. Schenker, R. B. Maxwell, R. Ganz, andM. Leunig. Arthroscopic management of femoroacetabular impingement:osteoplasty technique and literature review. The American journal of sportsmedicine, 35(9):1571?80, Sept. 2007. ISSN 1552-3365.doi:10.1177/0363546507300258. URL101http://www.ncbi.nlm.nih.gov/pubmed/17420508. ? pages 13, 19, 22, 23, 28,68, 69, 78C. Plaskos, A. J. Hodgson, K. Inkpen, and R. W. McGraw. Bone cutting errors intotal knee arthroplasty. The Journal of Arthroplasty, 17(6):698?705, Sept.2002. ISSN 08835403. doi:10.1054/arth.2002.33564. URLhttp://linkinghub.elsevier.com/retrieve/pii/S0883540302000736. ? pages 86J. Rand and D. Ilstrup. Survivorship analysis of total knee arthroplasty.Cumulative rates of survival of 9200 total knee arthroplasties. Journal of Bone& Joint Surgery Am, 73(3):397?409, 1991. ? pages 84G. Rivkin and M. Liebergall. Challenges of technology integration andcomputer-assisted surgery. The Journal of bone and joint surgery. Americanvolume, 91 Suppl 1:13?6, Feb. 2009. ISSN 1535-1386.doi:10.2106/JBJS.H.01410. URLhttp://www.ncbi.nlm.nih.gov/pubmed/19182015. ? pages 12T. G. Sampson. Complications of hip arthroscopy. Clinics in sports medicine, 20(4):831?5, Oct. 2001. ISSN 0278-5919. URLhttp://www.ncbi.nlm.nih.gov/pubmed/11675890. ? pages 13, 68T. G. Sampson. Arthroscopic treatment of femoroacetabular impingement.Techniques in Orthopaedics, 20(1):56?62, Dec. 2005. ISSN 1934-3418. URLhttp://www.ncbi.nlm.nih.gov/pubmed/19212569. ? pages 23, 68, 78T. G. Sampson, J. K. Nisbet, and J. M. Glick. Precision acromioplasty inarthroscopic subacromial decompression of the shoulder. Arthroscopy : thejournal of arthroscopic & related surgery : official publication of theArthroscopy Association of North America and the International ArthroscopyAssociation, 7(3):301?7, Jan. 1991. ISSN 0749-8063. URLhttp://www.ncbi.nlm.nih.gov/pubmed/1750941. ? pages xvii, 83, 86K. Satku. Unicompartmental knee arthroplasty: is it a step in the rightdirection??Surgical options for osteoarthritis of the knee. Singapore medicaljournal, 44(11):554?6, Nov. 2003. ISSN 0037-5675. URLhttp://www.ncbi.nlm.nih.gov/pubmed/15007492. ? pages 85J. M. Schimmels. Multidirectional compliance and constraint for improvedrobotic deburring. Part 1: improved positioning. Robotics andComputer-Integrated Manufacturing, 17(4):277?286, Aug. 2001a. ISSN07365845. doi:10.1016/S0736-5845(00)00059-4. URL102http://linkinghub.elsevier.com/retrieve/pii/S0736584500000594. ? pages xi, 7,8, 11, 12J. M. Schimmels. Multidirectional compliance and constraint for improvedrobotic deburring. Part 2: improved bracing. Robotics andComputer-Integrated Manufacturing, 17(4):287?294, Aug. 2001b. ISSN07365845. doi:10.1016/S0736-5845(00)00060-0. URLhttp://linkinghub.elsevier.com/retrieve/pii/S0736584500000600. ? pages 7, 11,12A. L. Simpson. The Computation and Visualization of Uncertainty in SurgicalNavigation. Doctor of philosophy, Queen?s University, 2010. ? pages 30J. R. Smith, P. E. Riches, and P. J. Rowe. Accuracy of a freehand sculpting toolfor unicondylar knee replacement. The international journal of medicalrobotics + computer assisted surgery : MRCAS, Online(June), 2013.doi:10.1002/rcs.1522. ? pages 25, 88, 89L. Solomon. Patterns of osteoarthritis of the hip. Journal of Bone & Joint Surgery,British Volume, 58(2):176?183, 1976. URLhttp://www.bjj.boneandjoint.org.uk/content/58-B/2/176.abstract. ? pages 15S. Stulberg and W. Harris. Acetabular dysplasia and development of osteoarthritisof the hip. In C. Mosby, editor, The Hip: Proceedings of the Second OpenScientific Meeting of the Hip Society, pages 82?93, St. Louis, 1974. ? pages 15S. Stulberg, L. Cordell, and W. Harris. Unrecognized childhood hip disease: amajor cause of idiopathic osteoarthritis of the hip. In C. Mosby, editor, TheHip: Proceedings of the Third Open Scientific Meeting of the Hip Society,pages 212?228, St. Louis, 1975. ? pages 15T. Stu?rmer, K. P. Gu?nther, and H. Brenner. Obesity, overweight and patterns ofosteoarthritis: the Ulm Osteoarthritis Study. Journal of clinical epidemiology,53(3):307?13, Mar. 2000. ISSN 0895-4356. URLhttp://www.ncbi.nlm.nih.gov/pubmed/10760642. ? pages 14O. B. Tabor Jr and O. B. Tabor. Unicompartmental Arthroplasty: a long-termfollow up study. J Arthroplasty, 13(4):373?379, 1998. ? pages 86R. Taylor, P. Jensen, L. Whitcomb, A. Barnes, R. Kumar, D. Stoianovici, P. Gupta,Z. Wang, E. Dejuan, and L. Kavoussi. A Steady-Hand Robotic System forMicrosurgical Augmentation. The International Journal of Robotics Research,18(12):1201?1210, Dec. 1999. ISSN 0278-3649.103doi:10.1177/02783649922067807. URLhttp://ijr.sagepub.com/cgi/doi/10.1177/02783649922067807. ? pages xi, 11,12M. a. Vitale, R. R. Arons, S. Hurwitz, C. S. Ahmad, and W. N. Levine. The risingincidence of acromioplasty. The Journal of bone and joint surgery. Americanvolume, 92(9):1842?50, Aug. 2010. ISSN 1535-1386.doi:10.2106/JBJS.I.01003. URLhttp://www.ncbi.nlm.nih.gov/pubmed/20686058. ? pages 83H. West. Kinematic analysis for the Design and Control of Braced Manipulators.Doctor of philosophy, Massachusetts Institute of Technology, 1986. ? pages 3,4, 6, 11S. C. Willimon, M. J. Philippon, and K. K. Briggs. Risk Factors for AdhesionsFollowing Hip Arthroscopy (SS-39). Arthroscopy: The Journal of Arthroscopic& Related Surgery, 27(5):e50?e51, May 2011. ISSN 07498063.doi:10.1016/j.arthro.2011.03.043. URLhttp://linkinghub.elsevier.com/retrieve/pii/S074980631100260X. ? pages 13,23, 68, 78L. P. Zebala, P. L. Schoenecker, and J. C. Clohisy. Anterior femoroacetabularimpingement: a diverse disease with evolving treatment options. The Iowaorthopaedic journal, 27:71?81, Jan. 2007. ISSN 1541-5457. URLhttp://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2150646&tool=pmcentrez&rendertype=abstract. ? pages 19J. Zupanc?ic? and T. Bajd. Comparison of position repeatability of a humanoperator and an industrial manipulating robot. Computers in biology andmedicine, 28(4):415?421, 1998. ISSN 0010-4825. URLhttp://www.sciencedirect.com/science/article/pii/S0010482598000195. ?pages xi, 7, 12104Appendix AVIVID 9i Laser ScannerSpecificationsTable A.1: VIVID 9i Laser Scanner Specifications. c?2006-2013 Konica Mi-nolta Sensing Americas, Inc.Attribute DescriptionType 3D Laser ScannerMeasuring Method Triangulation light block methodScan Range 0.6-1.0mLaser Scan Method Galvanometer-driven rotating mirrorLaser Class Class 2 (IEC60825-1), Class 1 (FDA)Accuracy (X,Y,Z) 0.05 mmPrecision (Z, ) 0.008 mmInput Time (per scan) 2.5 secAmbient Lighting Condition Office environment, 500 lx or lessRegulatory Approvals UL 61010A-1, CSA-C22.2 No.1010-1, etc.All information presented in this table was gathered using TELE lens at dis-tance of 0.6 m, with Field Calibration System, Konica Minoltas standard, at 20C.105Appendix BRepeatability AnalysesB.1 Scan-Rescan Validation for Proximal FemurSegmentsThe scan-rescan repeatability of the laser scanning workflow needed to be deter-mined to ensure that the sensitivity was small enough to detect meaningful differ-ences between samples.One proximal femur segment was scanned 10 different times using a Vivid9i laser scanner, with specifications listed in A, and 10 3D models were createdfrom the data in Rapidform XOR2. Each of the 10 models were registered to eachother and a mesh deviation was performed using Rapidform XOV2, as illustratedin Figure B.1. This allowed a total of 45 mesh deviation comparisons to be made,summarized in table B.1.The mean RMS error was 0.0286 +/- 0.00515mm. This gives an upper 95%error boundary of 0.0371mm. Chapter A lists the accuracy of the scanner as +/-0.05mm with the telephoto lens at 0.6m, with ambient office lighting. In our setupwe?re using the wide angle lens at a distance of approximately 0.6m, with muchlower ambient brightness, so our sensitivity of 0.0371mm is acceptable.106Table B.1: Raw data from the 45 mesh deviation comparisons made usingRapidform XOV2.Comparison RMS (mm) Std Dev. (mm) Comparison RMS (mm) Std Dev. (mm)Scan1 Scan2 0.0286 0.0275 Scan3 Scan10 0.0329 0.0327Scan1 Scan3 0.0276 0.0268 Scan4 Scan5 0.0177 0.0176Scan1 Scan4 0.0301 0.0295 Scan4 Scan6 0.0270 0.0235Scan1 Scan5 0.0254 0.0245 Scan4 Scan7 0.0234 0.0210Scan1 Scan6 0.0340 0.0270 Scan4 Scan8 0.0258 0.0245Scan1 Scan7 0.0296 0.0294 Scan4 Scan9 0.0189 0.0189Scan1 Scan8 0.0235 0.0235 Scan4 Scan10 0.0333 0.0331Scan1 Scan9 0.0273 0.0260 Scan5 Scan6 0.0252 0.0218Scan1 Scan10 0.0310 0.0307 Scan5 Scan7 0.0262 0.0236Scan2 Scan3 0.0253 0.0210 Scan5 Scan8 0.0233 0.0216Scan2 Scan4 0.0262 0.0223 Scan5 Scan9 0.0193 0.0193Scan2 Scan5 0.0245 0.0200 Scan5 Scan10 0.0315 0.0313Scan2 Scan6 0.0370 0.0243 Scan6 Scan7 0.0374 0.0286Scan2 Scan7 0.0226 0.0222 Scan6 Scan8 0.0337 0.0252Scan2 Scan8 0.0247 0.0238 Scan6 Scan9 0.0249 0.0214Scan2 Scan9 0.0267 0.0215 Scan6 Scan10 0.0337 0.0284Scan2 Scan10 0.0328 0.0310 Scan7 Scan8 0.0222 0.0222Scan3 Scan4 0.0206 0.0206 Scan7 Scan9 0.0220 0.0192Scan3 Scan5 0.0173 0.0173 Scan7 Scan10 0.0332 0.0327Scan3 Scan6 0.0264 0.0229 Scan8 Scan9 0.0247 0.0233Scan3 Scan7 0.0246 0.0221 Scan8 Scan10 0.0333 0.0331Scan3 Scan8 0.0242 0.0227 Scan9 Scan10 0.0292 0.0286Scan3 Scan9 0.0183 0.0183Figure B.1: Whole-deviation map from Rapidform XOV2107B.2 Scale RepeatabilityAn AL-500 scale from the Denver Instrument Company was used for all measuresof mass in this thesis. A proximal femur segment was weighed for 30 repetitionsto determine the accuracy of the scale, with results shown in table B.2. The meanfemur weight was 235.66 +/-0.0050g, a standard deviation smaller than the reso-lution of the scale. It was concluded that the scale had sufficient accuracy for ourstudy.Table B.2: Data from 30 repetitions of a single weight on a Denver InstrumentCompany AL-500 scale.Trial Weight (g) Trial Weight (g)1 235.65 16 235.642 235.66 17 235.643 235.66 18 235.654 235.66 19 235.665 235.66 20 235.656 235.66 21 235.657 235.65 22 235.658 235.66 23 235.649 235.66 24 235.6510 235.65 25 235.6511 235.65 26 235.6612 235.66 27 235.6513 235.66 28 235.6614 235.65 29 235.6515 235.65 30 235.65B.3 Scan-Rescan Validation for Wood SamplesThe scan-rescan repeatability of the revised laser scanning workflow, presentedin Section 3.3.1 needed to be determined to ensure that the sensitivity was smallenough to detect meaningful differences between samples, and for comparisonagainst our previous workflow.A scan was taken of a unmarked piece of wood with no sample on it and ana-108lyzed using the Rapidform workflow as previously described in Section 3.3.1. TheRMS value of a blank wood sample was 0.0283.A scan was taken of a piece of wood with milling marks on it and analyzedusing the Rapidform workflow as previously described in Section 3.3.1. In orderto assess the repeatability of the plane-fitting process, 20 different reference planeswere fit to the scan data, with the results summarized in B.3. A sample with millingmarks was used as the impact of such marks on the plane fitting process needed tobe assessed.Table B.3: RMS Errors from the 20 reference planes fit to a single sample forrepeatability analysis.Reference Plane RMS Error (mm) Reference Plane RMS Error (mm)1 0.3936 11 0.36472 0.3646 12 0.37513 0.3712 13 0.36624 0.3727 14 0.36995 0.3685 15 0.36826 0.3667 16 0.36467 0.3577 17 0.37228 0.3736 18 0.39319 0.3848 19 0.370410 0.3619 20 0.3772The mean RMS error was 0.372 +/- 0.0107mm. Two standard deviations, rep-resenting the variability of 95% of the data was 0.0214mm, which is less than theupper 95% error bound reported in Section B.1, so the process was deemed to besufficiently accurate.109Appendix CHam Hock Materials TestingSystemC.1 Characterizing Stiffness of Muscle TissueIn order to improve the fidelity of the testing system outlined in Chapter 2.3.4, a softtissue interface needed to be included. It was hypothesized and confirmed by ourclinical collaborator that arthroscopic tools receive support from the surroundingsoft tissue during the procedure, so their inclusion for future experimentation wasdeemed important.A 22lb porcine hock, shown in Figure C.1, was purchased from a local butcherand brought to room temperature to help with tissue mobility. At the time of ex-perimentation the sample was killed one day prior, preventing the effects of decay-related tissue stiffening. The hock was chosen as it provided a reasonable proxy forhuman thigh/gluteal tissue, including a skin-fat-fascia-muscle interface that arthro-scopic tools must travel through towards the hip joint.A material testing system (MTS), nicknamed the ham hock MTS, was createdusing Unistrut R?and pulleys which converted a downwards force from suspendedweights into a horizontal force connected to a hook placed into the tissue of interest,as shown in Figure C.2.Hooke?s law, F = kx, is a physical principle that states the force F needed toextend or compress a spring by a displacement x is proportional to that distance,110related through the stiffness k of the spring. We assume that the porcine tissue islinear-elastic, a clearly incorrect assumption given the highly hydrated nature of bi-ological materials, however this was done to allow us to determine an approximateorder of magnitude of support that arthroscopic tools receive in vivo. By makingthis assumption, we would only have to track the displacement of the hook, or aknown point on the static cord that connects the hook to the weights, under a knownforce.Displacement measurements were recorded using a video camera shown inFigure C.2. The skin and muscle were separately analysed using 5-15lb weights in5lb increments for a total of 3 trials per tissue type. The hook was inserted throughapproximately 6cm of muscle and fascia, a hypothesized minimum insertion depthfor an arthroscopy tool. Video footage was processed using Tracker, an open sourcevideo analysis and modeling tool ( c?2013 Douglas Brown), which would track apoint on the static cord during the loading phase of testing and provide time anddisplacement data.Results of the porcine muscle stiffness tests are shown in Figure C.3. Thestiffness value, determined using Hooke?s law, ranged from 3175.8 N/m to 5647.2N/m. This shows that the linear elastic assumption was incorrect, as hypothesized,however it gives us the desired order of magnitude (kN/m) of support that the toolsreceive during surgery. Results from the porcine skin tests were not included inthis analysis as it was decided that given the high mobility of the skin, it was notthe limiting factor in determining tissue stiffness.C.2 Foam Proxy of Muscle TissueWith the values of the porcine tissue stiffness determined, Figure C.3, a reasonableproxy needed to be determined. Three commercially available samples of foamwere tested:1. 2040 Foam2. 2050 Foam3. Blue Foam (sleeping pad)111Figure C.1: A 22lb pork hock purchased from a local butcher. The skin-fat-fascia-muscle interface is clearly visible from the side.Foam classification numbers refer to the density and resiliency of the foam,with a higher number offering a more dense/resilient foam.Foam samples were tested in a manner identical to section C.1.The 2040 foam results are illustrated in Figure C.4 and summarized in TableC.1. The 2050 foam results are illustrated in C.5 and summarized in Table C.2.The blue foam results are illustrated in C.6 and summarized in Table C.3.While the blue foam offered stiffness values that closely matched the measuredporcine tissue stiffness, it was commercially available in 9mm thickness only. Anapproximate thickness of 80mm was desired to provide a realistic amount of toolinsertion depth to mimic the thickness of a human thigh. Laminating the foam toachieve this thickness would cause the stiffness to become too high, akin to placingsprings in parallel and illustrated in C.7, so the 2050 foam was selected to be theproxy for soft tissue in future experimentation. It provided a balance between cost,112Figure C.2: The material testing system used to quantify the stiffness ofporcine tissue and foam proxies. It uses a system of pulleys to con-vert a downwards force into a horizontal one to apply to the sample.commercial availability, and variable thickness which best suited our needs.Table C.1: A comparison of the pork and 2040 foam stiffness under varyingloadsMass (lb) Pork Stiffness (N/m) 2040 Foam Stiffness (N/m)5 3175.8 1953.910 4170.6 2495.315 5647.2 2056.4113Figure C.3: Results from testing the stiffness of pork tissue. Time in secondsis on the x-axis, with displacement in meters on the y-axis. Accordingto Hooke?s law, under 5lbs the tissue stiffness was 3175.8 N/m, under10lbs the tissue stiffness was 4170.6 N/m, and under 15lbs the tissuestiffness was 5647.2 N/mTable C.2: A comparison of the pork and 2050 foam stiffness under varyingloadsMass (lb) Pork Stiffness (N/m) 2050 Foam Stiffness (N/m)5 3175.8 2482.610 4170.6 3045.615 5647.2 3399.2Table C.3: A comparison of the pork and 2050 foam stiffness under varyingloadsMass (lb) Pork Stiffness (N/m) Blue Foam Stiffness (N/m)5 3175.8 3249.110 4170.6 5265.615 5647.2 5223.4114Figure C.4: Results from testing the stiffness of 2040 foam. Time in secondsis on the x-axis, with displacement in meters on the y-axis. Accordingto Hooke?s law, under 5lbs the foam stiffness was 1953.9 N/m, under10lbs the foam stiffness was 2495.3 N/m, and under 15lbs the foamstiffness was 2056.4 N/m115Figure C.5: Results from testing the stiffness of 2050 foam. Time in secondsis on the x-axis, with displacement in meters on the y-axis. Accordingto Hooke?s law, under 5lbs the foam stiffness was 2482.6 N/m, under10lbs the foam stiffness was 3045.6 N/m, and under 15lbs the foamstiffness was 3399.2 N/m116Figure C.6: Results from testing the stiffness of blue foam. Time in secondsis on the x-axis, with displacement in meters on the y-axis. Accordingto Hooke?s law, under 5lbs the foam stiffness was 3249.1 N/m, under10lbs the foam stiffness was 5265.6 N/m, and under 15lbs the foamstiffness was 5223.4 N/m117Figure C.7: A visual comparison of the various foam properties when cor-rected for thickness, illustrating that the 2050 foam provided propertiesclosest to pork tissue.118Appendix DRCM + Spring Brace StiffnessOptimizationD.1 IntroductionThe spring used in Chapter 3 for the brace detailed in Section 3.9 was found ran-domly in a workshop and was used to test the theory that adding any axial stiffnessto a tool could improve performance. From 3.18, it was apparent that there wasa trend towards this hypothesis and more investigation into the effects of varyingspring stiffness on performance was warranted.D.2 MethodsFollowing the conclusion of the experiment in Chapter 3, a single subject who par-ticipated in the study continued to test the effect of spring stiffness on performance.In addition to the 1700 N/m spring used in Section 3.2.4, springs of 400 N/m, 2400N/m and 3000 N/m were tested. Spring constants were determined using a methodsimilar to Appendix C when manufacturer information was not available.The methods outlined in Section 3.3.1 were used to test three samples at eachspring stiffness by a single adult male subject (24 years, no surgical experience).119Figure D.1: Results comparing the impact of spring stiffness on time and ac-curacy of lesion removal.D.3 ResultsThe results of the testing outlined in Section D.2 are summarized in Figure D.1.In this instance, the control data consisted of that single subject?s Foam condi-tion data from Figure 3.18. There appears to be no relationship between stiffnessand performance, however the only spring stiffness to reduce both procedure lengthand error was the 3000 N/m spring.D.4 ConclusionsFor the purposes of this thesis, the most optimal spring, as tested on a single adultmale subject with no surgical experience, is one with a stiffness of 3000 N/m.120Appendix ELight Microscopy Analysis ofSpherical Burr WearDue to a shortage of commercially available 5.5mm abrader burrs for the Smith &Nephew handheld shaver used for this thesis, blades were used until the plastic in-terface near the handle melted, preventing the burr from spinning. During surgery,the tool would be used in combination with saline irrigation running through thehandle of the tool for cooling, however this wasn?t possible with our surgical setup.Instead, compressed air was fed through the tool and the plastic bearing interfacewas lubricated with lithium grease.While this would reduce the frequency with which the blades would melt, itstill didn?t prevent it. Blades would melt approximately every 3-4 subjects, beforeany noticeable decline in performance occurred. Without a standard measure forthe performance of a brand new blade versus one that had performed at least 48resections, we decided to inspect the cutting surface using light microscopy to seeif any visually detectable signs of dulling were present.A Carl Zeiss Axio Zoom. V16 stereo zoom microscope was used to inspect thecutting edges of a brand new 5.5mm abrader burr compared to the same burr thathad performed approximately 48 resections, shown in Figure E.1.Aside from the obvious impact of grease on the used burr, optical inspectionof the blades revealed no obvious defects that could impair cutting performance.We concluded that the blade changing schedule of approximately every 4 subjects121was of sufficient frequency to prevent any negative impact on the results from thestudied presented in this thesis.122Figure E.1: Light microscopy was used to compare the cutting surfaces of abrand new (top) and used (bottom) 5.5mm abrader burr.123Appendix FDebrief Questionnaire andResponsesF.1 Debrief Questionnaire124Page 1 of 1 Version: 02/04/2013 Advanced Tools for Computer-Assisted Orthopaedic Surgery: Effect of Instrument Bracing on Accuracy and Time of Femoral Neck Osteochondroplasty  Debrief Questionnaire   Please circle your response to each statement Statement SD D A SA I understood how to complete the task. 1 2 3 4 I understood how the test apparatus worked. 1 2 3 4 The EMG electrodes impaired my ability to use the tool. 1 2 3 4 The tool bracing was intuitive. 1 2 3 4 Tool bracing increased milling time. 1 2 3 4 Tool bracing increased control. 1 2 3 4 Tool bracing increased fatigue. 1 2 3 4 I enjoyed using the braced tool. 1 2 3 4  How could the bracing apparatus be improved? 1 ? SD Strongly Disagree 2 ? D Disagree 3 ? A Agree 4 ? SA Strongly Agree Subjects were asked to answer the questions posed in Section F.1 upon com-pletion of the study in Chapter 4, with answers summarized below.126Figure F.1: Pooled results (Surgeons and Non-Surgeons) from the debriefquestionnaire. Results were pooled due to the small Surgeon group(n=5) and the same trends being identified in each group.127

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074229/manifest

Comment

Related Items