Seminal vesicle intrafraction motion 69 sample Kolmogorov-Smirnov tests were used to determine whether or not the results for the left and right seminal vesicle belonged to similar distributions. All statistical analysis was performed using MATLAB version R2019a. In our previous work we reported the use of a Kalman filter in the registration pipeline.13 This Kalman filter was also implemented in this pipeline and allows flag-raising events in case sudden changes such as large intrafraction motion or possible mis-registrations occurred. All flagged events were visually inspected to determine if a mis-registration had occurred. Verification The obtained intrafraction motion results were verified by applying these results as a transformation to the original mask (as created from the daily anatomy). The transformed mask was then placed on the corresponding cine-MR dynamic. If the transformed mask was visually off from the position of the seminal vesicle in the cine-MR dynamic, the result was labelled as wrong. All datasets and results were manually and systematically verified. An example of a correctly transformed mask is visualised in Figure 1, image F. Figure 1 – Overview of cine-MR images from a single imaging session of one patient. Images A and C provide the transversal and sagittal slices at time point zero. Images B and D show the slices at the time point of 10 min. In these images, the influence of a gas pocket on the surrounding anatomy can be observed (arrow 1). Image E and F provide a sagittal close-up of the left seminal vesicle, with a delineation of the seminal vesicle in blue (arrow 2), and the found location by the tracking algorithm (cyan, arrow 3) in comparison with the original location (blue) in image F. The example provided in this figure showcases one of the largest seminal vesicle intrafraction translations seen in the dataset, with 9 mm anterior translation. 4
RkJQdWJsaXNoZXIy MTk4NDMw