The Use of 3D-Printed Phantoms for Evaluating CT Image Quality

Abstract

This thesis is about applying 3D printing in evaluating CT image quality and showing how this low-cost and fast-prototyping technique has the potential to create a paradigm shift in phantom design from simple geometrical objects to more sophisticated structures that incorporate anatomical features. Phantoms are used as surrogates for humans when evaluating the performance of CT imaging systems, but traditional molding techniques limit phantom designs in rigidly defined size, geometry, and material composition. Conventional phantoms are able to quantitatively evaluate the imaging system, but they overlook the complexity of the structural and textural properties of human anatomy. The high cost of the phantoms also limits the feasibility of customization and patient-specific individualization. CT image quality is highly dependent on various factors, such as the object’s size, geometry, the reconstruction method, and the image post-processing. Due to the object-dependence of CT image quality, it is important to evaluate on realistic objects. The anthropomorphic phantoms with known texture features are therefore needed to quantitative evaluate CT system. To this end, we utilize the recently emerged 3D-printing technology for constructing low-cost anthropomorphic phantoms. We devise a methodology for 3D-printed phantom inserts that represent the structural and textural properties that may be placed inside a large anthropomorphic phantom to simulate the bulk properties of the human body. We expect that this general method can be used as a repeatable and robust imaging performance assessment methodology that includes quantitative analysis of the reconstruction of specific textures and anthropomorphic features.

Keywords

3D-Printing, Phantom, Computed Tomography, Image Quality, Texture Reproduction

Read Thesis