Assessing geometric accuracy in MRI-based planning for stereotactic radiosurgery of brain tumours

Abstract

Objective: To conduct phantom-based assessments on two Magnetic Resonance Imaging (MRI) scanners to assess their static magnetic field (B0) inhomogeneity and separate sequence dependent from sequence independent distortions. The ultimate goal of this thesis is to ensure that the distortions in planning images used for Stereotactic Radiosurgery (SRS) treatment planning are less than 1 mm over a 150 mm diameter volume. Methods: The B0 inhomogeneity was assessed on the Burwood and Christchurch scanners using two methods with a spherical phantom; the bandwidth difference and phase difference map methods. A custom-built 3D phantom was used to separate sequence independent from sequence dependent distortions, and to measure the total distortion over three volumes in an MRI sequence that had been optimised for SRS requirements. Results: B0 inhomogeneity was lower on the Christchurch scanner versus the Burwood scanner at isocentre using the bandwidth difference method (0.5384 ppm vs. 0.8352 ppm), and at treatment position using the phase difference map method (1.1181 ppm vs. 1.7214 ppm). The phase difference map method at isocentre was lower on the Burwood scanner (0.5923 ppm) vs. the Christchurch scanner (0.6376 ppm). Inhomogeneity variance using standard shim produced lower results on the Christchurch over the Burwood scanner both at isocentre (0.16±0.03 mm vs. 0.23±0.02 mm) and treatment position (0.16±0.02 mm vs. 0.28±0.03 mm), however, tune up shim performed better on the Burwood scanner compared to the Christchurch scanner at isocentre (0.23±0.02 mm vs. 0.31±0.08 mm) and treatment position (0.24±0.02 mm vs. 0.29±0.03 mm). A sequence optimised for SRS planning produced distortions of less than 1 mm on the Christchurch scanner in both positions over all the volumes analysed. The Burwood scanner produced results less than 1 mm at isocentre but at treatment position the distortions increased to above 1 mm over the volumes analysed. Conclusion: The results shows that the Christchurch scanner with second order shim available and higher gradient strength produced lower B0 inhomogeneity in the treatment position, meeting the requirement of less than 1 mm distortion over a 150 mm diameter sphere. The smaller distortions produced with the Christchurch scanner make it suitable for use in SRS planning.