A new system for functional proton radiosurgery has been proposed. The goal of the system is to target specific brain areas with high doses of proton beams with submillimeter accuracy. High-energy proton beams have exquisitely sharp lateral penumbra and are, therefore, ideal for functional radiosurgery. Localizing the anatomical target with an MRI-based fiducial system requires correction of gradient nonlinearity distortions inherent in the scanner images. Modern MR scanners are particularly prone to such distortions due to wider bores and stronger gradient fields. The gradient nonlinearity correction described in this work is based on a high-resolution 3D MR scan of a cube phantom. Using a least-squares fitting procedure correction parameters are found that convert the geometrically warped planes of the cube into the ideal planes. In this paper, we describe the initial data processing and quality checks performed before the data is used for estimation of correction parameters. 1.

Abstract—Targeting of brain tumors or diseased brain tissues with neurosurgical instruments or radiosurgery beams requires magnetic resonance image (MRI) sets with high spatial resolution. In recent years, the spatial resolution of MRI scanners has increased considerably with the introduction of highfield (3T) technology. As a result, isotropic submillimeter voxel sizes are more commonly being used for medical procedures. On the other hand, gradient nonlinearities introduce geometric distortions and errors into the image data. As MRI technologies continue to progress, this problem has not been resolved. We have developed a gradient nonlinearity correction method based on a cubic phantom MRI data set. The approach utilizes a sum of spherical harmonics to model the geometrically warped planes of the cube. Once the polynomial parameters are known they can be applied to correct arbitrary image sets acquired with the same scanner. In this paper, we give a detailed description of the Matlab software performing the modeling and correction functions and discuss the possibility to accelerate the code using

A new system for functional proton radiosurgery has been proposed, and is currently under development at Loma Linda University Medical Center, Loma Linda, CA. The goal of the system is to target specific brain areas with high doses of proton beams with submillimeter accuracy. High-energy proton beams have exquisitely sharp lateral penumbra and are, therefore, ideal for functional radiosurgery. Localizing the anatomical target with an MRI-based fiducial system requires correction of gradient nonlinearity distortions inherent in the scanner images. Modern MR scanners are particularly prone to such distortions due to wider bores and stronger gradient fields. The gradient nonlinearity correction described in this work is based on a high-resolution 3D MR scan of a cube phantom. Using a least-squares fitting procedure correction parameters are found that convert the geometrically warped planes of the cube into the ideal planes. The accuracy of the correction method is then tested and verified using a second phantom containing targeting markers. The locations of the markers reported by