OSR Journal of Student Research


The current process of intracranial radiosurgery treatment uses implanted titanium fiducials in the skull to assist in alignment of the patient. These fiducials add an element of physical and emotional stress to the patient, and scheduling the implantation procedures adds a delay of a few extra days before the radiosurgery procedure can begin. During the radiosurgery treatment, each proton beam is manually aligned by the therapist/physician with X-ray images and the fiducials that are visible on these images. This method of alignment can be time-intensive and requires personnel who are specifically trained in patient alignment. We propose a new method using image registration to automate this process in an effort to eliminate the need for surgical implantation of fiducials prior to treatment as well as to improve the accuracy and efficiency of alignment during treatment. Image registration is a technique used to align a moving image with respect to its known fixed image. Several methods of image registration are used for comparison: an enhanced correlation coefficient maximization algorithm, a mutual information maximization algorithm, and an extended phase correlation algorithm. Accuracy, robustness, and performance are emphasized in the comparison of these algorithms. Due to patient privacy, test images from MATLAB will be shown in this paper. This research was conducted under the clinical supervision of Dr. Andrew Wroe and Dr. Reinhard Schulte of the Loma Linda University Medical Center (LLUMC).