Open Collections

Javanmardi, Arash

Rapid advancements in high-throughput computing have used medical images to compute quantitative and minable data that have large implications on clinical events in oncology management. This has generated a new area of research referred to as “Radiomics”. The extraction of quantitative imaging features and textures from medical images provides unique data about tissue pathology that allows for better diagnosis and treatment planning for cancer patients. Limitations of radiomics arise when different scanners show inconsistent radiomic features due to the intrinsic differences between scanners. The conventional method of image quality validation is done using the NEMA Image Quality phantom, where 18FFDG is inserted into fillable spheres. This method ignores the heterogeneity and anatomy of real tumours. Therefore it is incapable of verifying and validating crucial radiomic features capable of providing highly accurate diagnosis and prognosis of cancer. The aim of this thesis is to harmonize cross-center PET scanners and validate radiomic features using a novel 3D-printed phantom that displays heterogeneous activity distributions. Digital Reference Objects (DROs) with known radiomic features were used as the reference objects for the 3D-printing of the variable density phantom and testing of radiomic features [1]. Using the in-house made PhantomCreator MATLAB function, 3D printable DRO phantoms were generated.

Text

University of British Columbia. PHYS 449

Vancouver : University of British Columbia Library

10.14288/1.0396965

Science, Faculty of; Physics and Astronomy, Department of

Undergraduate

DSpace