Registration guided simulation of prostate movement for radiation therapy

Developing an accurate dosage plan for radiation therapy is crucial to the success of the treatment, yet many obstacles still exist in spite of the application of the modern technology. One such challenge is revealed in registering CT scans and MRI images, that capture the location and geometry of the prostate. Due to bodily functions and breathing of the patients, the prostate moves with respect to different images. Relying solely on computer vision techniques such as the demons algorithm, can at best give a quantitative measurement of the differences between images. How the differences are created remains unknown. We develop a pipeline in which we first employ simulations of the movement and deformation of the rectum to depict its influence on the position variation of the prostate, then calculate the displacement fields using demons image registration algorithm, and lastly use the displacement field to correct the simulations to match with the target image. Potentially, physicians can better devise dosage planning knowing how the prostate moves during the treatment.