Room: Karl Dean Ballroom C
Purpose: Evaluate scanner performance and compare task-based image quality on ViewRayâ€™s 0.345T MR integrated treatment delivery system to that of typical diagnostic MR scanners.
Methods: Monthly ACR phantom T1 scans were acquired on a ViewRay 0.345T MR system over 1.5 years. We used twenty-five signal averages and adjusted receiver bandwidth to increase SNR to that typical of a 1.5T scanner to imitate ACR testing. Performance measures based on these scans were gathered and evaluated, with performance measures from 8 diagnostic systems (5 1.5T, 3 3T) over 5-7 years as a reference. T1-weighted pre-treatment liver MR scans were acquired on both diagnostic and ViewRay systems for 8 patients. Liver lesions and liver were segmented and approved by a radiation oncologist. Lesion contrast-to-noise ratio (CNR) was evaluated to understand the impact of scanner and protocol on pre-treatment lesion definition.
Results: Phantom-based geometric accuracy, slice positioning, and spatial resolution, essential on ViewRay systems, were within ACR tolerances and had typical mean performance and variability compared to diagnostic scanners. Low-contrast detectability was lower and more variable (Âµ=28.0, Ïƒ=3.6) than diagnostic LCD measures (Âµ=38.6, Ïƒ=1.4 overall) despite attempted SNR recovery. Percent signal ghosting was highly variable and outside ACR tolerances on several occasions, and uniformity was also relatively variable on the ViewRay. Slice thickness was outside ACR tolerances in 8/19 monthly measurements on the ViewRay, always greater than expected (>5.7mm). Liver lesion SNR was sufficient for detectability, though CNR was highly variable on both diagnostic and ViewRay systems, with no significant difference observed.
Conclusion: Variations in performance measures (LCD, ghosting) on the ViewRay system were associated with the lower SNR caused by lower field strength, though no significant difference in liver lesion CNR was observed. Imaging performance of integrated MR systems should be considered carefully, particularly when implementing automated image processing for treatment planning.