Automatic Deformable Registration on Prostate Cine-MRI Images for Studying Intra-fraction Motion in Supine and Prone Position with and Without Rectal Balloon

Prior studies of intra-fraction motion had relied on points to define prostate position. They include X-ray, Calypso electromagnetic markers or point of interests (POIs) in cine-MRI images. Defining the prostate position by points provide a surrogate for the prostate position. We present the first volume analysis of prostate the intra-fraction motion for patients at different settings with or without a rectal balloon (WRB) and without (WORB). A total of 68 Cine-MRI studies were done in 17 different series with 4 scans per series, consisting of supine and prone, each WRB and WORB. Each san was performed for 4 minutes; the time is close to complete one field of proton treatment. Furthermore, it is the maximum time before the chemical shift to distort the spatial resolution of images due to the short excitation time between pulses (240 ms). By employing automatic deformable registration developed by ViewRay Inc., the prostate and pelvis structures were segmented. The accuracy of automatic segmentation was carefully inspected. A program was then developed to obtain centroid of prostate from the prostate outline. Prostate position of each image was evaluated in reference to the initial image. After inspecting automatic segmentation, no correction was needed for prostate outline. The mean prostate position over 240 seconds of all series was small: supine WRB 0.217 mm, WORB 0.397 mm and prone WRB 0.267 mm, WORB 0.318 mm, although smaller for the supine patients WRB (p < 0.0001). However, the variation of the prostate position during 240s was larger: supine WRB 1.01 mm, WORB 1.20 mm and prone WRB 1.47 mm, WORB 2.144 mm (p < 0.001). The margins needed to cover all patients 100% of the treatments during 95% of the single beam time (4 min) were: supine WRB 2.21 mm, WORB 2.75 mm and prone WRB 3.13 mm, WORB 4.53 mm (p < 0.001). A strong relation was seen between time and prostate motion: the prostate will remain within 2.0 mm and 3.2 min, respectively, in 2 and 4 minutes for the best case, supine WRB, and within 4.2 mm and 5.3 mm for the worse case, prone WORB. The probabilities for prostate staying within +/− 1 mm to its initial position are: 94% supine WRB x-axis (i.e., anterior-posterior) and 88% y-axis (i.e., superior-inferior); supine WORB 78% x-axis and 75% for y-axis; prone WORB 67% x-axis and 78% axis; and prone WORB 55% x-axis and 83% y-axis. After employing the image guidance (either cone-beam or X-rays) to setup patient, intra-fraction motion becomes the largest error and needs to be included into the treatment margin. Reducing the time between after setup and the finalization of treatment or beam will require smaller intra-fraction margin. Prostate position for supine WRB Cine-MRI scans was very stable and can use a small treatment margin.