UBC Faculty Research and Publications

Accuracy of 177Lu activity quantification in SPECT imaging: a phantom study Uribe, Carlos F; Esquinas, Pedro L; Tanguay, Jesse; Gonzalez, Marjorie; Gaudin, Emilie; Beauregard, Jean-Mathieu; Celler, Anna Jan 7, 2017

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

Item Metadata

Download

Media
52383-40658_2016_Article_170.pdf [ 2.81MB ]
Metadata
JSON: 52383-1.0368611.json
JSON-LD: 52383-1.0368611-ld.json
RDF/XML (Pretty): 52383-1.0368611-rdf.xml
RDF/JSON: 52383-1.0368611-rdf.json
Turtle: 52383-1.0368611-turtle.txt
N-Triples: 52383-1.0368611-rdf-ntriples.txt
Original Record: 52383-1.0368611-source.json
Full Text
52383-1.0368611-fulltext.txt
Citation
52383-1.0368611.ris

Full Text

ORIGINAL RESEARCH Open AccessAccuracy of 177SPECT imaginCarlos F. Uribe1,2,3*, Pedro L. EJean-Mathieu Beauregard6,7 a* Correspondence: curibe@bccrc.ca1Medical Imaging Research Group,Department of Radiology, Universityof British Columbia, Vancouver,British Colombia, Canada2Department of Physics andAstronomy, University of BritishColumbia, Vancouver, BritishColombia, CanadaFull list of author information isavailable at the end of the articlenon-uniform density.nEJNMMI PhysicsUribe et al. EJNMMI Physics  (2017) 4:2 DOI 10.1186/s40658-016-0170-3developed.Keywords: 177Lu, Quantification, Dosimetry, TEW, APDI, Scatter correctioConclusions: Following the MIRD recommendations for data acquisition andreconstruction resulted in accurate activity quantification (errors <5% for largeobjects). However, techniques for better organ/tumor segmentation must still be©LpiLu activity quantification ing: a phantom studysquinas1,2, Jesse Tanguay1, Marjorie Gonzalez4, Emilie Gaudin5,nd Anna Celler1AbstractBackground: The aim of the study is to assess accuracy of activity quantificationof 177Lu studies performed according to recommendations provided by thecommittee on Medical Internal Radiation Dose (MIRD) pamphlets 23 and 26. Theperformances of two scatter correction and three segmentation methods werecompared. Additionally, the accuracy of tomographic and planar methods fordetermination of the camera normalization factor (CNF) was evaluated.Eight phantoms containing inserts of different sizes and shapes placed in air,water, and radioactive background were scanned using a Siemens SymbiaTSPECT/CT camera. Planar and tomographic scans with 177Lu sources were used tomeasure CNF. Images were reconstructed with our SPEQToR software using resolutionrecovery, attenuation, and two scatter correction methods (analytical photon distributioninterpolated (APDI) and triple energy window (TEW)). Segmentation was performedusing a fixed threshold method for both air and cold water scans. For hot waterexperiments three segmentation methods were compared as folows: a 40% fixedthreshold, segmentation based on CT images, and our iterative adaptive dualthresholding (IADT). Quantification error, defined as the percent difference betweenexperimental and true activities, was evaluated.Results: Quantification error for scans in air was better for TEW scatter correction(<6%) than for APDI (<11%). This trend was reversed for scans in water (<10% forAPDI and <14% for TEW). For hot water, the best results (<18% for small objectsand <5% for objects >100 ml) were obtained when APDI and IADT were used forscatter correction and segmentation, respectively. Additionally, we showed thatplanar acquisitions with scatter correction and tomographic scans provide similarCNF values. This is an important finding because planar acquisitions are easier toperform than tomographic scans. TEW and APDI resulted in similar quantificationerrors with APDI showing a small advantage for objects placed in medium withThe Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 Internationalicense (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,rovided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, andndicate if changes were made.Uribe et al. EJNMMI Physics  (2017) 4:2 Page 2 of 20BackgroundTargeted radionuclide therapy (TRT) uses pharmaceuticals labeled with radioisotopeswhich emit particles (β’s or α’s) to deliver dose directly to tumors, while avoiding irradiatinghealthy tissues. This approach has been used in management of a number of oncologicaland other disorders [1, 2]. In particular, TRT has been shown to produce very encouragingresults in treatment of neuroendocrine tumors (NETs) using 177Lu labeled peptide analogs[3]. Unfortunately, treatment outcomes vary greatly between patients and cure remainsrare. There is growing evidence that this lack of consistent response may be due to the“one-size-fits-all” treatment approach, where all patients are injected with the same,relatively low activity of approximately 7400 MBq/cycle [4]. However, radiotracer uptakein tumors and healthy tissue varies greatly between patients, which results in a big groupof under-treated individuals [5, 6]. It is generally believed that treatment plans using injec-tions based on individualized dose assessments could significantly improve TRT out-comes; therefore, they should become routine, as is already the case in external beamtherapies [7, 8].This opinion prompted the recent publication of a series of committee on Medical In-ternal Radiation Dose (MIRD) pamphlets specifying guidelines for SPECT-based activityquantification which is necessary for personalized, image-based dosimetry for TRT. Thesepublications clearly identify the sequence of procedures which have to be performed andcorrections which have to be applied to generate quantitative images of activity distributionin the patient body and to obtain information about how this activity changes over time.The general recommendations are outlined in MIRD pamphlet 23 [9], while MIRDpamphlet 26 [10] focuses specifically on studies performed with 177Lu.However, these documents do not provide information about the accuracy of activityquantification that can be achieved when following these guidelines. Additionally, insome cases, the recommended procedures or corrections are not uniquely defined andthe final selection of the method is left to the user. Consequently, the decision regardingwhich method should be used may be difficult, especially if the accuracies of differentapproaches are not uniquely characterized.Previous studies investigating the accuracy of 177Lu quantification used a number ofdifferent phantom configurations, and different data acquisition and processingmethods. For example, Beauregard et al. [11] measured activity of 175 ml cylindersplaced in a large cylindrical phantom and reported 6.6 ± 3.5% average errors. They usedcamera manufacturer’s image reconstruction software with attenuation and dual energywindow (DEW) [12] scatter corrections, while contributions from scattered high-energy photons were ignored. Our group's previous 177Lu quantification studies [13],based on the 113 keV photopeak using low-energy high-resolution (LEHR) collimatorand the analytical photon distribution interpolated (APDI) scatter correction method[14, 15], resulted in 2% accuracy error for a 70 ml container scanned in air and in water(without background activity). In parallel, seven scatter correction methods and differentcollimators were evaluated by de Nijs et al. [16] using phantom containing hot spheres inwarm background with concentration ratio close to 13:1. Although triple energy window(TEW) [17] was found to not be suitable when imaging the 113 keV photopeak, for the208 keV peak TEW, or even DEW method, resulted in activity estimation errors close to10% for the largest 37 ml sphere. Sanders et al. [18] scanned a cylinder uniformly filledwith activity containing small spheres (0.5–16 ml) and used TEW scatter correction.Uribe et al. EJNMMI Physics  (2017) 4:2 Page 3 of 20The resulting quantification errors were about 20% for the largest sphere. Most recently,Hippeläinen et al. [19] tested the accuracy of quantification using spheres filled with 177Luplaced inside a torso phantom. The ratio of spheres to background activity concentrationswas very high (30:1). The best quantification was achieved when attenuation, collimatorresponse, and Monte Carlo-based scatter corrections were implemented in the recon-struction algorithm, and 15% error was reported for the largest sphere.Once quantitatively accurate images are reconstructed, the counts in each voxel mustbe translated into activity values [9, 10]. This is done using experimentally determinedcamera normalization factor (CNF), and both planar and tomographic methods havebeen proposed for this purpose. While planar acquisitions are easier and faster to per-form than tomographic scans, it is believed that they can only be used when highlyaccurate scatter and attenuation corrections are included in image reconstructions.Tomographic acquisitions are said to better approximate scatter and attenuation inpatients resulting in better compensation for possible errors in the quantitative recon-structions [9]. We decided to verify these claims in a challenging case of 177Lu.Finally, once the images of quantitative activity distributions are available, to performorgan dosimetry (for example, following OLINDA protocol [20]), the exact informationabout organ of interest activity and its mass are required. Considering this task, wecompared the accuracy of activity quantification for objects scanned in hot background.Three different segmentation approaches were evaluated: (i) a 40% fixed thresholdmethod which is commonly used in clinical studies [21], (ii) segmentation based ontrue organ volume obtained directly from CT images, and (iii) our iterative adaptivedual thresholding (IADT) method [22].To summarize, using a large series of phantom experiments, we investigated thefollowing four questions:1. What accuracy of 177Lu activity quantification can be achieved in phantomexperiments performed following recommendations of the MIRD pamphlet 26?2. What is the accuracy of quantification for the two scatter correction methods:(i)the fast and easy, but approximate TEW method [17], and (ii) a rigorous, butcomputationally intensive APDI [14, 15] method.3. What is the best method to measure the camera normalization factor (CNF),and what are the uncertainties associated with each of these methods?4. How is the accuracy of activity quantification in an organ affected by the employedsegmentation method?MethodsTo answer the questions listed in the previous section, we performed a series of experi-ments using a wide range of phantom geometries and different attenuation and scatterconditions.1. Phantoms with hot inserts placed in air. The purpose of scanning hot objectsof different shapes and sizes placed in air was to evaluate the accuracy of ourreconstructions in situations with relatively small amount of attenuation and scatter.2. Phantoms with hot inserts placed in water. Hot objects (with activity added) placedin cylinders filled with cold water (no activity added) allowed us to compare theperformance of the two scatter correction methods. In these experiments phantomswith uniform (Jaszczak cylinder) and non-uniform (thorax phantom) density distri-butions were used.3. Phantoms with hot inserts in warm water (containing activity). The purpose ofThe experimental details and the procedures are explained in the following sections.energy windows: the 208 keV photopeak window (PW), lower scatter window (LSW), andrized in Table 2.Uribe et al. EJNMMI Physics  (2017) 4:2 Page 4 of 20Jaszczak phantom with spherical insertsSeven spheres with volumes ranging from 0.5 to 113 ml filled with water containing3.19 ± 0.17 MBq/ml of 177Lu were placed inside the cylindrical Jaszczak phantom (DataSpectrum Corporation, NC, USA; 22.2 cm diameter, 19.5 cm height). The phantomwas first scanned with the spheres in air (referred to as Jaszczak spheres in air). Next,the cylinder was filled with water (no activity), (referred to as Jaszczak spheres in coldwater), and a second scan was performed.A third scan (referred to as Jaszczak spheres in warm water) was performed withactivity present in the background. The concentration of activity in the backgroundwas equal to 0.49 ± 0.03 MBq/ml, resulting in signal-to-background ratio (SBR) of 6.4.Table 1 Energy window used in the 177Lu phantom experimentsName Lower limit [keV] Center [keV] Upper limit [keV]LSW 153.0 170.0 187.0upper scatter window (USW) (see Table 1).Phantom experimentsTo test the quantification accuracy of our methods, the first series of experiments used eightdifferent phantom configurations. Photos of the phantoms (described below) are shown inFig. 1. The volumes and activities of each of the inserts and acquisition times are summa-Data acquisitionsAll scans were performed using a SymbiaT (Siemens Medical, Germany) SPECT/CTcamerawith a medium-energy low-penetration (MELP) collimator. For each tomographic scan 90projections were acquired with a non-circular orbit using a 128 x 128 matrix and threeexperiments with activity in the background was to test the performance of thethree different segmentation approaches in determination of the activities and thevolumes of objects of different sizes. The configuration with hot objects in warmbackground best approximates the conditions present in patient studies.4. Phantoms for CNF and IADT calibrations. The objective of these additionalexperiments (performed separately from the series discussed above) was to (a)measure the camera normalization factor using images from tomographicacquisitions of objects with different levels of activity, and (b) determinethresholding curves for the IADT segmentation method.PW 187.2 208.0 228.8USW 229.5 255.0 280.2Uribe et al. EJNMMI Physics  (2017) 4:2 Page 5 of 20Jaszczak with cylindrical bottle insertsNext, to evaluate the potential dependence of the accuracy of quantification on the shapeof the object, two of the previous scans were repeated with cylindrical inserts. Two8.5 mL (B1 and B2), two 12 ml (B3 and B4), and one 16 ml (B5) bottles were filled withactivity with concentration of 5.11 ± 0.22 MBq/ml. Similar to spherical inserts study, thephantom was first scanned in air (Jaszczak: bottles in air). A second scan (Jaszczak: bottlesin warm water) was performed with the phantom filled with water containing concentra-tion of activity of 0.104 ± 0.003 MBq/ml. The SBR in this case was equal to 49.1.Thorax phantomThe aim of the third experiment was to determine the accuracy of 177Lu quantifica-tion in challenging non-uniform attenuation conditions. Four identical bottles(T1–T4) of 34 ml volume were filled with the activity shown in Table 2.The bottles were attached in different locations inside the thorax phantom (EllipticalLung-Spine Body Phantom, Data Spectrum Corporation) as follows: Bottle T1 was attached to the spine. Bottle T2 was placed under one of the lungs. Bottle T3 was placed in between the lungs and the spine.Fig. 1 Photos of the phantoms used in the experimentsaccuracy of quantificationB1 8.5 42.6Uribe et al. EJNMMI Physics  (2017) 4:2 Page 6 of 20Jaszczak bottles in air 20 B2 8.5 42.2B3 12. 61.3Jaszczak bottles in warm water 20 B4 12 60.0B5 16 90.1B6 34 90.1Thorax phantom in cold water 20 T1 34 300.0T2 34 302.9Phantom insertsPhantom configuration Projection duration [s] Name Volume [ml] True activity [MBq]Jaszczak spheres in air 20 S0 0.5 1.7S1 1.0 3.2S2 2.0 6.1Jaszczak spheres in cold water 30 S3 4.0 12.3S4 8.0 24.5S5 16 48.3Jaszczak spheres in warm water 30 S6 113 351.6Table 2 Summary of the phantoms which were used in the experiments performed to test the Bottle T4 was attached to a beef bone placed at the inferior part of the phantom.This bone was used to model a tumor close to an object having true boneattenuation (spine insert only approximates real bone density) and to make theattenuation at the bottom of the phantom less homogeneous.The phantom was filled with cold water (no activity) and scanned.Bottles on the camera bedThe final experimental series aimed to determine the accuracy of quantification in the mostdifficult case. Four bottles filled with activity were placed between irregularly shaped objects(water bags) creating non-uniform attenuation and scatter conditions. Bottles C1, C2, C3,and C4 were filled with activity (Table 2), and the first scan was performed with bottlesplaced in air, directly on the camera bed, bottles on bed. Then, the bottles were placedbetween four 2 L water bags (two below and two on top of the bottles) and a second scanwas performed, bottles on bed with water bags.Jaszczak with changing SBR activity concentrationsBesides the eight phantom configurations described above, five additional, separateexperiments were performed to determine the calibration curves for our IADT method.T3 34 302.9T4 34 302.9Bottles on bed 10 C1 34 —C2 63 854.0Bottles on bed with water bags 10 C3 182 846.5C4 199 1182.6In these experiments, six bottles (V = 17.0 to 199.5 ml) filled with 177Lu activity withconcentration of 0.62 ± 0.03 MBq/ml placed inside a Jaszczak phantom were scanned.Five SPECT/CT scans were performed: for the first scan the phantom was empty sothe inserts were scanned in air and for the following four scans, the activity in the largecylinder (background) was gradually increased to obtain four different SBR’s (14.0, 8.0,5.3, and 4.1). The first three of these scans were also used to determine CNF (discussedin the next section) in tomographic mode to be compared with planar scans.Planar and tomographic acquisitions for CNF determinationOur third objective was to evaluate the accuracy of different methods that can be usedto determine the CNF. To this end, we performed several experiments with 177Lu usinga MELP collimator. Table 3 summarizes these experiments.All scans, except for B (Table 3), were performed at the Nuclear Medicine Department,Vancouver General Hospital (Vancouver, British Colombia) and used three energywindows. The uniform cylinder (scan D) filled with 177Lu-chloride was scanned at theL’Hôtel-Dieu de Québec site of the CHU de Québec–Université Laval center (Quebec City,Table 3 Details of planar and tomographic acquisitions performed to determine the CNF for 177LuMode (phantom) Fig. 5 data Phantom Source info Acquisition parametersA Planara (point source) Planar methods1 and 2Activity 11.7 MBqVolume 0.5 mlScan duration = 10 minCollimator MELPPW [187.2–228.8] keVLSW [153.0–187.0] keVUSW [229.5–280.2] keVB Tomographicb (uniformlyhot cylinder)SPECT cylinder Total activity659.6 MBqVolume 6500 ml#Projections = 96Time/projection = 10sUribe et al. EJNMMI Physics  (2017) 4:2 Page 7 of 20E Tomographic (6 spheresin warm water)SPECT HW2 Total activity 681.3 MBqVol bottles [17–199.5] ml#Projections = 90Time/projection = 30saThe camera sensitivity does not depend on the source-collimator distance although increase in recorded counts mayoccur due to septal penetration if the sources are placed very close to the collimator. To minimize this effect, in thisC Tomographic (6 bottlesin air)SPECT air Total activity233.4 MBqVol bottles [17–199.5] ml#Projections = 90Time/projection = 20sD Tomographic (6 spheresin warm water)SPECT HW1 Total activity 489.1 MBqVol bottles [17–199.5] ml#Projections = 90Time/projection = 30sstudy the point sources were placed at 30 cm from the collimator surfacebUniform cylinder was scanned at the L’Hôtel-Dieu de Québec site of the CHU de Québec–Université Laval center(Quebec City, Canada) and dual energy window was performed for scatter correction on this phantomUribe et al. EJNMMI Physics  (2017) 4:2 Page 8 of 20QC). It used dual energy window (DEW) scatter correction: PW (187.2–228.8 keV) andLSW (166.4–187.2 keV) instead of three windows.Image reconstructionReconstructions were performed using our in-house developed software incorporatedinto a graphical user interface (GUI) named single photon emission quantitative tomo-graphic reconstruction (SPEQToR). It allows the user to select the reconstruction algo-rithm and the corrections to be included in this reconstruction. All reconstructionsused the standard ordered subsets expectation maximization (OSEM) algorithm (10 sub-sets/6 iterations) with CT-based attenuation correction (AC), resolution recovery (RR),and scatter correction (SC).The number of subsets and iterations applied in the OSEM algorithm were selectedbased on a small study performed prior to the one described here. Because in patientsthe sizes of lesions are not known a priori, the chosen number of 10 subsets and 6 iter-ations seem to provide a good compromise between quantification accuracy for big andsmall lesions and between accuracy and speed of the reconstruction. Moreover, al-though it is important to accurately quantify small tumors, kidneys remain the limitingorgan in Lu-177 therapies.Our resolution recovery method incorporates the distance-dependent geometricresponse of the detector-collimator system (using Gaussian functions) directly into thesystem matrix. The model does not include septal penetration or scatter. The details ofthe resolution recovery method are described in Blinder et al. [23].In imaging studies using 177Lu, besides primary photons, the 208 keV photopeakwindow contains Bremsstrahlung, self-scatter, and high-energy scatter components.Bremsstrahlung for 177Lu (188Re and 90Y) has been already investigated by us and wasshown to be negligible (<0.02%) [24]. Although the contribution of high-energy photonsalso is low in the USW, our simulations indicated that it is not negligible (3%, the numberof photons in the PW). Most of the high-energy photons scatter within the crystal of thecamera after passing through the collimator. Therefore, better accuracy is expected whenthese high-energy photons are corrected. In summary, only corrections for the two scattercomponents have been incorporated into the OSEM formula, but Bremsstrahlung emis-sions were neglected:Xlþ1j ¼XljXi∈ΩiCijXi∈ΩiCijY iXkCijXlk þ S þ Hð1ÞIn this equation, Yi represents the measured projections, Cij is the system matrix inwhich attenuation and resolution recovery information are included, and Xlj is the lthestimate of the image. The terms S and H in the denominator represent the self-scatterand high-energy scatter components, respectively.The two scatter correction methods which were investigated in this study representthe two very different approaches: APDI is an analytical scatter correction method thatcombines the first estimate of activity distribution in the object with information aboutits density distribution (attenuation map) with the Klein-Nishina formula to determineS, the contribution of scattered photons to the measured projections. Subsequently, theS component, together with H (the high-energy scatter component which is calculatedUribe et al. EJNMMI Physics  (2017) 4:2 Page 9 of 20prim sthe photopeak window Cpw.Cprim ¼ Cpw−Cs ð4ÞThe CNF was calculated using equation similar to Eq. (2), but with primary photonsonly.CNF ¼ CprimA tdð Þ ð5ÞFor the four tomographic scans (B, C, D, and E in Table 3), the CNF was determined bytaking the total number of counts in each reconstructed image (Crec) and dividing themby the activity in the phantom multiplied by the total scan duration which was given bythe number of projections (np) multiplied by the duration of each projection (tp).CNF ¼ CrecA tp  np  ð6ÞThe reconstructions of the SPECT/CT data were performed using the methodsdescribed in section 2.2, with TEW scatter correction.SegmentationMethod 2. Using the counts in the PW corrected for scatter with the TEW method,the scattered counts (Cs) were estimated using the counts in the LSW (Cls), the USW(Cus), and the widths of each window wls, wus, and wpw for the LSW, USW, and PW,respectively.Cs ¼ Clswls þCuswus wpw2ð3ÞThe primary photons (C ) were then determined by subtracting C from counts inof the scan (td).CNF ¼ CpwA td ð2ÞDetermination of the CNFFor the planar scan (A in Table 3), two methods were tested when determining CNF:Method 1. Using all counts in the PW, the CNF was calculated by dividing the totalnumber of counts in the image (Cpw) by the activity of the source (A) and the durationusing counts recorded in USW) are incorporated into the denominator of Eq. (1). Themethod has been shown to provide accurate quantification for 99mTc [14, 25, 26],188Re [27], and 177Lu [13], but is computationally intensive.TEW, on the other hand, is fast and has been implemented on many SPECT/CT camerasmaking it widely available in Nuclear Medicine departments. Its further advantage is the factthat counts recorded in LSW and USW provide a joint estimate of S +H without any needfor additional calculations. It is, however, an approximate method that does not model thespatial distribution of scatter [10].No smoothing was applied to any energy window in either APDI or TEW method.All segmentations were performed in three dimensions (3D) and consisted of two steps.First, to separate each of the analyzed objects from other objects in the field of viewFor scan in cold water, the threshold had to be raised to account for residual scatteredUribe et al. EJNMMI Physics  (2017) 4:2 Page 10 of 20counts in the background. For these scans a 1% fixed threshold was applied to the datain the large VOI. The threshold values were determined using the mean of the 9 voxelscontaining the highest activity values within the initially selected ROIs.These low-level thresholds were chosen to include counts which due to spill out orpartial volume effects are found outside the physical boundary of the object. For scanswith warm background, the accuracy of quantification was compared for the threesegmentation methods:1. A fixed 40% threshold.2. Segmentation based on CT images.3. Our iterative adaptive dual thresholding (IADT) method [22].A fixed threshold method is often used for segmentations performed in clinical studieswith threshold values usually close to 40% [21]. Therefore, this 40% fixed threshold wasused here to serve as a reference for both cases of SBR = 6.4 and SBR = 49.1. Next, for theCT-based segmentation, the volumes of interest were defined by segmenting the CTimage, so that for each object, its segmented volume was equal to its physical volume, asshown on the CT.Finally, the IADT method was used. The IADT method [22] uses two differentthresholding curves (one for activity and a different one for volume) to iteratively deter-mine values of the two thresholds which are optimized for the SBR measured in thesegmented VOI. There is no need for any a priori information about the target volumeor SBR. The calibration curves must be determined in a separate calibration experi-ment. In our case, acquisitions of the Jaszczak phantom with inserts with four differentSBR activity concentrations (section 2.2.5) were used. As described by Grimes et al.[22], the data from these experiments were reconstructed twice, and two separate setsof thresholding calibration curves (one set for TEW and one for APDI) were obtained.These curves were used for the iterative segmentation of the current phantom experi-ments with a hot background activity.Quantification accuracy and statistical analysisThe accuracy of activity quantification in each object was estimated for all phantomconfigurations, scatter corrections, and segmentation methods. The numerical value ofthe error of quantification was calculated as the percentage difference of the objectactivity determined using each of the analyzed methods relative to the “true” activity asmeasured in the dose calibrator. According to the dose calibrator manufacturer(FOV), a rough segmentation of the phantom was performed by manually drawing alarge region of interest (ROI) around the desired object in each slice. These ROIswere then combined to create a large 3D volume of interest (VOI) around the objectwhich was then used to perform a finer segmentation using one of the methodsdescribed below.For scans in air, a fixed threshold of 0.1% was applied to the data in the large VOI.(Capintec), the variability of these measurements is equal to 5%. Box plots were gener-ated for each phantom configuration and for both scatter correction methods.the null hypothesis that the two scatter correction methods provide the sameaccuracy of activity quantification (the hypothesis was accepted for p < 0.05). Thiscloser to the true values of activity. For scans performed in cold water, the activity inUribe et al. EJNMMI Physics  (2017) 4:2 Page 11 of 20volumes smaller or equal to 2 ml was underestimated by both scatter correctionmethods, while for larger volume reconstructions with APDI outperformed those donewith TEW. This effect was especially noticeable for bottles scanned in the non-uniformthorax phantom. In particular, bottle T4, which was attached to the bone,showed excellent quantification accuracy for APDI. Finally, for the large bottles(160–200 ml) placed on the camera bed and surrounded by the water bags, bothscatter correction methods behaved very similarly, providing very good accuracyof activity quantification.Determination of the CNFFigure 5 shows the normalization factors obtained with both planar and tomographicscans of 177Lu sources. The values obtained from all tomographic acquisitions agreewas done separately for each phantom configuration, and in the case of hot waterfor the three segmentation methods.Additionally, to compare the three segmentation methods used in the warm waterphantom configurations, a Mood’s median test was applied. This test allows for the com-parison of the three segmentation methods at once. In this case, the null hypothesisassumed that the segmentation methods provided the same results (the hypothesiswas accepted for p < 0.05).ResultsPhantom experimentsFigures 2, 3, and 4 summarize the results of the analysis of activity quantification in ob-jects of different sizes and shapes placed in air, cold water, and warm water,respectively. In the top part of the figures, the quantification errors for individual ob-jects, sorted by increasing volumes from the left (S1) to the right (C4), are presented.The distributions of accuracy errors for both scatter corrections and, in case of scans inwarm water—for the three segmentation methods are shown in the lower part of eachfigure. In all cases, the CNF values obtained using the planar method 2 was used. Pleasenote that the following objects were not included in our analysis: (a) sphere S0 wasexcluded due to its very small size, (b) bottle C1 was not analyzed because we dis-covered that a large error was made in determination of its activity, and (c) sphereS1 was not analyzed because it was not visible in images reconstructed from scanswith hot water.In general, for phantoms scanned in air and reconstructed with APDI scatter correc-tion method, the activities were overestimated by up to 11%, while TEW results wereA statistical analysis of the errors for images reconstructed using TEW andAPDI scatter corrections methods was performed. As the distributions of theseerrors showed to be non-Gaussian, the Mann-Whitney test was applied to checkwithin 7% with each other and with the values obtained from planar scan, providingthat the TEW scatter correction was applied (i.e., method 2).Uribe et al. EJNMMI Physics  (2017) 4:2 Page 12 of 20SegmentationFigures 4a, b, and c compare the accuracy of quantification achieved with the three seg-mentation methods for objects scanned in the hot background. The objects have beensorted by volumes from the smallest object (S2) to the largest one (S6). Both 40% fixedthreshold and CT-based segmentation methods grossly underestimated the activity forboth TEW and APDI scatter correction methods. The IADT segmentation method seemedto be the most effective in recovering the true activity of each object. For small volumes(<17 ml), due to spill out caused by partial volume effects (PVE), the activity estimates werenot very accurate (>10%). For objects larger than 34 ml errors below, 10% were achieved.Discussion177Lu quantificationFor objects placed in air (Fig. 2), the effects of attenuation and scatter were small and mostlydue to attenuation and scatter within the object itself. Additionally, segmentation usingfixed threshold with a very low value of 0.1% allowed us to recover the majority of activityFig. 2 Quantification errors for phantom inserts with different shapes and volumes scanned in air (a), anderror distribution for both scatter correction methods, TEW and APDI (b). The horizontal dashed lines in (a)mark the range (maximum and minimum) of the deviations from the truth. The boxes in (b) represent therange of variation (interquartile range-IQR) of the distributionsUribe et al. EJNMMI Physics  (2017) 4:2 Page 13 of 20in each object. Therefore, in general, the accuracy of activity quantification achieved in theseexperiments was very good; activities reconstructed with TEW scatter correction differedfor most of the objects by less than 5% from the true values. These results confirmed thatour reconstruction method perform well in low scatter and attenuation conditions.It is interesting to note that for scans in air, the results obtained from thereconstructions with TEW were better than those from the APDI method, whichin all cases overestimated the activities. This effect, especially strong for smallerobjects, can be explained by the fact that although APDI accurately models scat-ter distributions, it calculates scatter using attenuation maps of the object withlarge (interpolated) voxels. Large PVE at the boundary between water and aircaused APDI method to underestimate scatter contributions.On the other hand, when bottles were placed in air directly on the camera bed (C2, C3,and C4), although the results from APDI still overestimated the activity, its accuracy wasbetter than TEW, with errors lower than 3%. In this case, the volumes were larger, so PVEwas relatively small and non-uniform scatter conditions (due to close proximity of thecamera bed) were modeled better by APDI than by the TEW method.Fig. 3 Quantification errors for phantom inserts with different shapes and volumes scanned in cold waterand error distribution (b) for both scatter correction methods, TEW and APDI (b). The horizontal dashed linesin (a) mark the maximum and minimum deviations from the truth. The boxes in (b) represent the range ofvariation (interquartile range-IQR) of the distributionsUribe et al. EJNMMI Physics  (2017) 4:2 Page 14 of 20Fig. 4 Quantification errors (with uncertainties) for phantom inserts with different shapes and volumesFor scans performed with cold water background (no activity), the reconstructions withTEW scatter correction underestimated the activity in small objects and overestimated itin large objects. Furthermore, the reconstructions with APDI provided more accurateactivity values and also qualitatively sharper images than those reconstructed with TEW.In the thorax phantom (non-uniform attenuation case), both scatter correction methodsscanned in warm water and segmented with three different methods: 40% fixed threshold (a), CT based (b),and IADT (c). The error distribution for both scatter correction methods, TEW and APDI (d). The horizontallines in (a–c) mark the maximum and minimum deviations from the truth. The boxes in (d) represent therange of variation (interquartile range-IQR) of the distributionsFig. 5 Camera normalization factors (CNF) 177Lu obtained using different methods. The horizontal dashed linerepresents the average value as determined from the planar method 2 and the tomographic acquisitionsUribe et al. EJNMMI Physics  (2017) 4:2 Page 15 of 20overestimated the activity of the inserts (T1, T2, T3, and T4), but again, APDI was moreor even much more accurate than TEW. In particular, activities in bottle T3 (which wasplaced between the lungs and spine) and in bottle T4 (which was placed in contact withthe beef bone) were both highly overestimated by TEW while APDI reconstructed thetrue activity very accurately (error <1%). Finally, for the three large bottles placed betweenwater bags (C2, C3, and C4), both scatter correction methods performed equally well.Figures 2 and 3 do not show a big difference between quantification errors of smallobjects and big objects. Usually when fixed threshold values are chosen, the effect ofpartial volume effects is larger for small objects making their activity quantificationworse than for large objects. However, the fixed thresholds used in this study made thepartial volume effects negligible.Figures 2b, 3b, and 4d show box plots summarizing quantification errors for differentphantom scans. These plots are especially useful when comparing the performance ofAPDI and TEW scatter correction methods. In clinics, lesions of different sizes andlocations might be present in the same patient and quantification accuracy in thesecases will vary. Therefore, it is important to see the range of these variations and dis-persion of the accuracy which can be achieved with the two scatter correction methods,to identify potential advantages of each method. Smaller variations around the meansuggest a better overall performance as long as quantification remains within a few per-cent from the truth.For the scans in air (Fig. 2b) and cold water (Fig. 3b), the Mann-Whitney test showedthat distributions of TEW and APDI were significantly different. In particular, for thescans in air, the mean quantification error in images reconstructed with TEW was closeto 1% of the true value. However, the distribution of errors was skewed towards nega-tive values, meaning that TEW was likely to underestimate the activity. This effect wasespecially pronounced in scans with large bottles (C1, C2, and C3) placed directly onthe camera bed, as TEW performed better when the objects are further from the scat-tering medium. As discussed before, for the scans in air, the reconstructions with APDIin general overestimated the activity values, but produced more accurate results in anon-uniform attenuation environment.For the scans in cold water, the box plot (Fig. 3b) shows that the distributions oferrors for both TEW and APDI were skewed towards lower values (the mean is lowerthan the line representing the median). While reconstructions with TEW tended tooverestimate the activity, those performed with APDI mostly underestimated it. Onaverage, the APDI results were more accurate than TEW and the error variation aboutthe mean value was smaller.The situation became more challenging when warm background was present. In thiscase, sphere S1 had been removed from the analysis because, due to its small volume(1 ml), it could not be distinguished from the background. For the same reason, thequantification results for sphere S2 (2 ml), although reported here, may not be trusted.For the scans in warm water (Fig. 4d), the Mann-Whitney test did not show statisti-cally significant differences between the two scatter correction methods when data wereanalyzed using the same segmentation algorithm.Our analysis of the three segmentation methods showed that activity in the objectsdetermined using the 40% threshold, which is often used in clinical studies [21], wasgrossly underestimated. This threshold value clearly was too high and did not properly~40%). However, the CT-based segmentation method may be difficult to use in patientUribe et al. EJNMMI Physics  (2017) 4:2 Page 16 of 20studies because boundaries of organs/tumors are not always visible in CT images. Oneway to deal with this problem is to apply recovery coefficients, as was done in the studyby Ilan et al. [28]. They scanned a known phantom and used a 42% fixed threshold forsegmentation. Then, for each object size, they calculated recovery coefficient to correctfor missing activity.In general, the best results were obtained with the IADT segmentation method. Inparticular, for volumes equal or larger than 34 ml, the accuracy of quantification wasbetter than 5%. Good performance of the IADT method is at least partly due its “adaptive”character, i.e., the thresholding curves that were used in segmentation have been obtainedfrom phantom scans reconstructed with the same method (TEW and APDI) as theimages which were segmented. Please note, however, that the IADT method failswhen activity distribution in the analyzed VOI is non-uniform as then the SBR cannotbe correctly determined (see [22]). Therefore, the quality of each object’s segmentationmust be critically evaluated by the operator.Finally, the box plot of Fig. 4d summarizes the results for the three segmentation andtwo scatter correction methods. Because all experiments with warm backgroundinvolved phantoms with uniform distribution of attenuating medium, the quantificationaccuracy of both scatter correction methods was very similar for all three segmentationapproaches. Nevertheless, all three segmentation methods underestimate the activityindicating the need for improvement in segmentation techniques. The Mood’s mediantest showed that the three methods (i.e., 40%, CT, and IADT segmentations) weresignificantly different for the samples reconstructed using the two scatter correctionmethods.Additionally, if imaged activities are very high, the count losses caused by cameradead time must be corrected for. In our phantom experiments the activities were rela-tively low, so no dead time corrections were necessary. Nevertheless, this topic hasbeen extensively investigated by us in a separate study and will be reported shortly.Camera normalizationAlthough methods to determine camera normalization factor have been extensivelyinvestigated, there is no consensus which approach is the best. Dewaraja et al. [29] usedMonte Carlo simulations of 131I activity to compare two approaches: a planar scan of apoint source and two tomographic scans of a phantom uniformly filled with activityand the same phantom filled with water with a hot sphere placed at the center. Theyfound that the camera normalization obtained using hot sphere placed inside theaccount for the spill out effect. Segmentations which used the true dimensions of eachobject (based on CT images) were strongly affected by PVEs, especially for the smallestspheres (S2–S4). The accuracy obtained for small spheres was similar to that achievedwith the 40% threshold (errors ~40–60%), but for larger objects, the CT-based methodperformed substantially better than the 40% threshold. Interestingly, results were betterfor small cylindrical objects (B1 and B2 bottles, 8.5 ml volume, errors <15%)) than forthe spheres having approximately the same volumes (S4 sphere, 8 ml volume, errorwarm-water-filled phantom provided better quantification than the uniform phantom.However, if sufficiently large volume of interest was used, accuracies of quantificationUribe et al. EJNMMI Physics  (2017) 4:2 Page 17 of 20obtained from the point source and water cylinder were similar. Ljungberg et al. [30]compared simulations and experimentally measured CNF using a planar scans of the131I point source. When counts in the whole field of view (FOV) of the camera weretaken into account, good agreement was obtained between simulations and measure-ment. Zeintl et al. [31] used a cylinder uniformly filled with 99mTc and reported CNFvalues similar to those obtained by our group using planar scans of point sources [26].Frey et al. [32] suggests that the use of a phantom tomographic acquisition, instead of aplanar scan of a small source, is the optimal approach. Lastly, D’Arienzo et al. [33] per-formed tomographic scans of point sources in air and extended phantoms and showedbetter quantification accuracy for CNF determined from phantom studies than frompoint sources. They attribute this result to the fact that attenuation and scatter correc-tions were not incorporated into the reconstruction of the point source.The results of our CNF determination are presented in Fig. 5. In general, the valuesof normalization determined using the planar method 2 (with scatter correction) andall tomographic scans were very similar. However, when scatter correction was notapplied (method 1), the CNF values were overestimated.Also, the CNF values obtained from the planar method 2 and the tomographic scansagree well, and the maximum deviation from the mean remains below 5%. Due to con-tribution from scatter of high-energy photons, planar method 1 produced higher CNFvalue than the other methods.Although our experiments were performed on a Siemens SymbiaT camera, ourexperience indicates that similar results should be expected for data collected usingother modern cameras. It is important, however, that CNF is correctly determined. Ourresults suggest that planar scans of point sources placed in air are sufficient for thedetermination of the camera calibration factor. However, for planar scans, scatter correc-tion should be applied to remove contributions from background and scattered photons.This is particularly important for 177Lu and other isotopes in which background from highenergy scattered photons can be substantial.ConclusionsThe accuracy of activity determination was evaluated for 177Lu imaging studies performedaccording to MIRD Pamphlet 23 and 26 recommendations. Several phantoms containinginserts with a variety of shapes and sizes placed in different attenuation conditions wereperformed.Our results showed that for phantoms scanned in air, errors in activity quantificationbelow 6.5 and 11.5% were achieved for large (i.e., >100 ml) and small (i.e., <100 ml)objects, respectively, for both TEW and APDI scatter correction methods. For phantomsscanned in cold water, APDI provided activities within 2% of the true values for volumes>100 ml, and within 10% for smaller objects. While APDI performed better in challengingsituations involving non-uniform attenuating medium, for scans of phantoms withuniform density, the performance of both scatter correction methods was very similar,indicating that much faster TEW method can be sufficient when scanning body areaswith relatively uniform tissue distributions.For scans with activity added to the phantom’s background, all three segmentationmethods (40% fixed threshold, CT-based, and our IADT method) underestimated theactivity in the inserts. However, our quantification results for scans in air and coldUribe et al. EJNMMI Physics  (2017) 4:2 Page 18 of 20water suggest that our modeling of scatter and attenuation are sufficiently accurate.However, segmentation is still a challenge. For both TEW and APDI scatter correctionmethods, IADT segmentation provided the best quantification accuracy. For IADT, thefact that only small differences were seen between the two scatter correction algorithmscould be attributed to the fact that the threshold curves which were used to segmentobjects were created using reconstructions with the same sets of corrections as thoseused in the phantom reconstructions.Regarding the camera calibration, our results showed that, when TEW correctionsare applied, the CNF values determined using planar scans of the point sources werethe same as those from the tomographic scans of extended sources. This is an importantfinding, considering that planar scans of point sources are much easier to perform thantomographic acquisitions of extended phantoms.In summary, although the exact activity quantification (thus also dose calculation)of very small lesions still remains challenging, IADT segmentation performed well,within 5%, for larger inserts (>34 ml), thus it would be appropriate for use in dosimetrycalculation of medium to large organs, like the kidneys. Finally, for lesions located inregions with non-uniform attenuation, the APDI scatter correction method would berecommended.The results of this work strongly suggest that in patient scans, similarly good quanti-fication accuracy can be achieved in determination of activity in larger organs and/ortumors. This is an important conclusion, as the exact quantification of activity in anorgan is crucial for the accuracy of dosimetry calculation for this organ and such dosimetryis particularly important for critical organs, such as, for example, the kidneys.Our results indicate that the methods used in this study allow for accurate determinationof 177Lu activity, but also suggest that organ/tumor segmentation still remains challenging.Further efforts should be made in developing better segmentation algorithms for nuclearmedicine data.Additional fileAdditional file 1: Table S1. Quantification error data for Figs. 2 and 3, inserts in air and water. Table S2.Quantification error data for Fig. 4, inserts in hot water. (DOC 67 kb)AbbreviationsAC: Attenuation correction; APDI: Analytical photon distribution interpolated; CNF: Camera normalization factor;DEW: Double energy window; FOV: Field of view; IADT: Interative adaptive dual thresholding; LEHR: Low energyhigh resolution; LSW: Lower scatter window; MELP: Medium energy low penetration; MIRD: Committee on medicalinternal radiation dose; NETs: Neuroendocrine tumors; OSEM: Ordered subsets expectation maximization; PVE: Partialvolume effects; PW: Photopeak window; ROI: Region of interest; RR: Resolution recovery; SBR: Signal-to-backgroundratio; SC: Scatter correction; TEW: Triple energy window; TRT: Targeted radionuclide therapy; USW: Upper scatterwindow; VOI: Volume of interestFundingThis project has been funded by the Natural Sciences and Engineering Research Council of Canada, awardnumber 194512-12.Availability of data and materialsThe dataset(s) supporting the conclusions of this article is (are) included within the article (and its Additional file 1.Authors’ contributionsCU was responsible for the acquisitions, reconstructions, segmentations, and data analysis and participated in thestudy design with AC. PE, JT, MG, and AC assisted with the acquisitions and analysis of data. EG and JMB assistedwith the acquisitions in Quebec, Canada. All authors read and approved the manuscript.bia,ec–15. Vandervoort E, Celler A, Wells G, Blinder S, Dixon K, Member S, et al. Implementation of an analytically basedpubmedcentral.nih.gov/articlerender.fcgi?artid=3969156&tool=pmcentrez&rendertype=abstract.Uribe et al. EJNMMI Physics  (2017) 4:2 Page 19 of 2017. Ogawa K, Harata Y, Ichihara T, Kubo A, Hashimoto S. A practical method for position-dependent Compton-scattercorrection in single photon emission CT. IEEE Trans Med Imaging. 1991;10:408–12. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18222843.scatter correction in SPECT reconstructions. IEEE Trans Nucl Sci. 2005;52:645–53.16. de Nijs R, Lagerburg V, Klausen TL, Holm S. Improving quantitative dosimetry in 177Lu-DOTATATE SPECT by energywindow-based scatter corrections. Nucl Med Commun. 2014;35:522–33. Available from: http://www.8. Strigari L, Konijnenberg M, Chiesa C, Bardies M, Du Y, Gleisner KS, et al. The evidence base for the use of internaldosimetry in the clinical practice of molecular radiotherapy. Eur J Nucl Med Mol Imaging. 2014;41:1976–88.9. Dewaraja YK, Frey EC, Sgouros G, Brill AB, Roberson P, Zanzonico PB, et al. MIRD pamphlet no. 23: quantitativeSPECT for patient-specific 3-dimensional dosimetry in internal radionuclide therapy. J Nucl Med. 2012;53:1310–25.Available from: http://www.ncbi.nlm.nih.gov/pubmed/22743252.10. Ljungberg M, Celler A, Konijnenberg MW, Eckerman KF, Dewaraja YK, Sjogreen Gleisner K. MIRD pamphlet no. 26:joint EANM/MIRD guidelines for quantitative 177Lu SPECT applied for dosimetry of radiopharmaceutical therapy.J. Nucl. Med. [Internet]. 2015;9881. Available from: https://www.ncbi.nlm.nih.gov/pubmed/26471692.11. Beauregard J-M, Hofman MS, Pereira JM, Eu P, Hicks RJ. Quantitative 177Lu SPECT (QSPECT) imaging using acommercially available SPECT/CT system. Cancer Imaging. 2011;11:56–66. Available from: https://www.ncbi.nlm.nih.gov/pubmed/21684829.12. Jaszczak RJ, Greer KL, Floyd CE, Harris CC, Coleman RE, Jaszczak J, et al. Improved SPECT quantification usingcompensation for scattered photons. J Nucl Med. 1984;25:893–900.13. Shcherbinin S, Piwowarska-Bilska H, Celler A, Birkenfeld B. Quantitative SPECT/CT reconstruction for 177Lu and177Lu/90Y targeted radionuclide therapies. Phys Med Biol. 2012;57:5733–47. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22948135.14. Wells RGG, Celler A, Harrop R. Analytical calculation of photon distributions in SPECT projections. IEEE Trans NuclSci. 1998;45:3202–14. Available from: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=736199.Competing interestsThe authors declare that they have no competing interests.Consent for publicationNot applicable.Ethics approval and consent to participateNot applicable.Author details1Medical Imaging Research Group, Department of Radiology, University of British Columbia, Vancouver, BritishColombia, Canada. 2Department of Physics and Astronomy, University of British Columbia, Vancouver, British ColomCanada. 3Department of Molecular Oncology, BC Cancer Research Centre, Vancouver, British Colombia, Canada.4Vancouver Coastal Health Authority, Vancouver, British Colombia, Canada. 5Department of Physics, EngineeringPhysics and Optics, Université Laval, Quebec City, Quebec, Canada. 6Department of Medical Imaging, CHU de QuebUniversité Laval, Quebec City, Quebec, Canada. 7Department of Radiology and Nuclear Medicine, Université Laval,Quebec City, Quebec, Canada.Received: 2 November 2016 Accepted: 20 December 2016References1. Dash A, Knapp FF (Russ), Pillai MRA. Targeted radionuclide therapy—an overview. Curr. Radiopharm. [Internet]. 2013;6:152–80. Available from: https://www.ncbi.nlm.nih.gov/pubmed/24059327.2. Ersahin D, Doddamane I, Cheng D. Targeted radionuclide therapy. Cancers (Basel). 2011;3:3838–55. Available from:http://www.mdpi.com/2072-6694/3/4/3838.3. Kashyap R, Hofman MS, Michael M, Kong G, Akhurst T, Eu P, et al. Favourable outcomes of 177Lu-octreotatepeptide receptor chemoradionuclide therapy in patients with FDG-avid neuroendocrine tumours. Eur J Nucl Med.2015;42:176–85.4. Radojewski P, Dumont R, Marincek N, Brunner P, Mäcke HR, Müller-Brand J, et al. Towards tailoredradiopeptide therapy. Eur J Nucl Med Mol Imaging. 2015;42:1231–7. Available from: https://www.ncbi.nlm.nih.gov/pubmed/25792454.5. Beauregard J-M, Hofman MS, Kong G, Hicks RJ. The tumour sink effect on the biodistribution of 68Ga-DOTA-octreotate: implications for peptide receptor radionuclide therapy. Eur J Nucl Med Mol Imaging. 2012;39:50–6.Available from: http://www.ncbi.nlm.nih.gov/pubmed/21932117.6. Sjogreen Gleisner K, Brolin G, Sundlov A, Mjekiqi E, Ostlund K, Tennvall J, et al. Long-term retention of 177Lu/177mLu-Dotatate in patients investigated by gamma spectrometry and gamma camera imaging. J Nucl Med. 2015;56:976–85. Available from: https://www.ncbi.nlm.nih.gov/pubmed/25999429.7. Stabin M. The case for patient-specific dosimetry in radionuclide therapy. Cancer Biother. Radiopharm. [Internet].2008 [cited 2012 Apr 5];23. Available from: http://online.liebertpub.com/doi/abs/10.1089/cbr.2007.0445.18. Sanders JC, Kuwert T, Hornegger J, Ritt P. Quantitative SPECT/CT imaging of 177Lu with in vivo validation inpatients undergoing peptide receptor radionuclide therapy. Mol. Imaging Biol. [Internet]. 2014; Available from:https://www.ncbi.nlm.nih.gov/pubmed/25475521.from: http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1009254.24. Uribe CF, Esquinas PL, Gonzalez M, Celler A. Characteristics of Bremsstrahlung emissions of 177Lu, 188Re, and 90Yfor SPECT/CT quantification in radionuclide therapy. Phys Medica. 2016;32:691–700. Available from: https://www.ncbi.nlm.nih.gov/pubmed/27157626.25. Vandervoort E, Celler A, Harrop R. Implementation of an iterative scatter correction, the influence of attenuationmap quality and their effect on absolute quantitation in SPECT. Phys Med Biol. 2007;52:1527–45. Available from:http://www.ncbi.nlm.nih.gov/pubmed/17301469.26. Shcherbinin S, Celler A, Belhocine T, Vanderwerf R, Driedger A. Accuracy of quantitative reconstructions in SPECT/CT imaging. Phys Med Biol. 2008;53:4595–604.27. Shcherbinin S, Grimes J, Bator A, Cwikla JB, Celler A. Three-dimensional personalized dosimetry for 188Re liverselective internal radiation therapy based on quantitative post-treatment SPECT studies. Phys Med Biol. 2013;59:119–34. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24334821.Uribe et al. EJNMMI Physics  (2017) 4:2 Page 20 of 2028. Ilan E, Sandström M, Wassberg C, Sundin A, Garske-Román U, Eriksson B, et al. Dose response of pancreaticneuroendocrine tumours treated with peptide receptor radionuclide therapy using 177Lu-DOTATATE. J. Nucl. Med.[Internet]. 2015 [cited 2016 Dec 5];177–82. Available from: http://jnm.snmjournals.org/content/early/2015/01/15/jnumed.114.148437.abstract.29. Dewaraja YK, Wilderman SJ, Ljungberg M, Koral KF, Zasadny K, Kaminiski MS. Accurate dosimetry in 131Iradionuclide therapy using patient-specific, 3-dimensional methods for SPECT reconstruction and absorbed dosecalculation. J Nucl Med. 2005;46:840–9.30. Ljungberg M, Sjögreen K, Liu X, Frey E, Dewaraja Y, Strand S-E. A 3-dimensional absorbed dose calculationmethod based on quantitative SPECT for radionuclide therapy: evaluation for (131)I using Monte Carlo simulation.J Nucl Med. 2002;43:1101–9.31. Zeintl J, Vija AH, Yahil A, Hornegger J, Kuwert T. Quantitative accuracy of clinical 99mTc SPECT/CT using ordered-subsetexpectation maximization with 3-dimensional resolution recovery, attenuation, and scatter correction. J Nucl Med. 2010;51:921–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20484423.32. Frey EC, Humm JL, Ljungberg M. Accuracy and precision of radioactivity quantification in nuclear medicineimages. Semin Nucl Med. 2013;42:208–18.33. D’Arienzo M, Cazzato M, Cozzella ML, Cox M, D’Andrea M, Fazio A, et al. Gamma camera calibration and validationfor quantitative SPECT imaging with 177Lu. Appl Radiat Isot. 2016;112:156–64. Available from: http://www.sciencedirect.com/science/article/pii/S0969804316300902.Submit your manuscript to a journal and benefi t from:7 Convenient online submission7 Rigorous peer review7 Immediate publication on acceptance7 Open access: articles freely available online7 High visibility within the fi eld7 Retaining the copyright to your article19. Hippeläinen E, Tenhunen M, Mäenpää H, Sohlberg A. Quantitative accuracy of 177Lu SPECT reconstruction usingdifferent compensation methods: phantom and patient studies. EJNMMI Res. 2016;6:16. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4759452&tool=pmcentrez&rendertype=abstract.20. Stabin MG, Sparks RB, Crowe E. OLINDA/EXM: the second-generation personal computer software for internaldose assessment in nuclear medicine. J Nucl Med. 2005;46:1023–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15937315.21. Erdi YE, Wessels BW, Loew MH, Erdi AK. Threshold estimation in single photon emission computed tomographyand planar imaging for clinical radioimmunotherapy. Cancer Res. 1995;55:5823s–6s.22. Grimes J, Celler A, Shcherbinin S, Piwowarska-Bilska H, Birkenfeld B. The accuracy and reproducibility of SPECTtarget volumes and activities estimated using an iterative adaptive thresholding technique. Nucl Med Commun.2012;33:1254–66. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23010981.23. Blinder S, Celler A, Wells RG, Thomson D, Harrop R. Experimental verification of 3D detector responsecompensation using the OSEM reconstruction method. IEEE Nucl Sci Symp Conf Rec. 2001;4:2174–8. Available    Submit your next manuscript at 7 springeropen.com

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.52383.1-0368611/manifest

Comment

Related Items