In this thesis the surface reconstruction problem for radiation therapy is studied. The knowledge of the body's surface in the affected area is required to design and deliver proper radiation treatment for cancer patients. We develop an inexpensive non-contact optical 3D surface scanner prototype. It can easily be mounted on an x-ray treatment machine integrating unobtrusively into the current workflow.
We start with preliminary information on radiation therapy, explaining how the patient's body surface is used in planning and treatment, and discussing the currently available methods. Then, we review the commonly used optical 3D surface measurement methods, and discuss the advantages of structured light techniques for the stated problem. In our method, the projected pattern is the regular rectangular grid of colour-coded circular spots. Images of the light pattern deformed on the surface are photographed by an off-the-shelf inexpensive camera and analyzed by the developed software.
Finally, we present the special purpose image processing and surface reconstruction algorithm, implemented as a MATLAB package, and provide the experimental results. Optimization procedures are used in the pattern recognition code to process low-quality images efficiently. The software package can run on any standard PC and takes a couple of minutes to process one image.