Determining the 3D Structure of Bones Using Neural Networks on X-Ray Images

Academic Advisor: Nicolas Lehmann , Agnès Voisard , Peter Böttcher
Degree: Master of Science (M.Sc.)
Status: assigned

Abstract

CT (Computer Tomography) is an imaging procedure used to create 3-dimensional images of certain body regions. These images assist doctors by providing essential information about their patients, which is relevant to their further treatment. This procedure, however, exposes patients to high doses of radiation. This drawback motivates us to look for alternative, less harmful methods that can nevertheless provide relevant 3D information about a body region.

The goal of this thesis is to develop such a method. Specifically, neural networks should be used to match 2-dimensional X-ray (Roentgen) images to their corresponding 3D structures. An X-ray scan is much less harmful to a patient than a CT scan.

Hint: Knowledge of neural networks and image processing are desirable, but not necessary.