Part I | Chapter 4 76 An example of the intrafraction motion paths due to a gas pocket is visualised as the black traces in Figure 3. In this figure, it can be observed that a gas pocket appears near the time point of 4-5 min, leading to sudden translation displacements. The motion paths in the left-right translation graph show that both seminal vesicles were displaced laterally, in addition to displacement in the anterior and cranial translation direction. Significant rotation about the left-right axis can be observed, combined with rotation about the anterior-posterior rotation axis due to rectal expansion. Sixteen cine-MR dynamics were removed due to mis-registrations. All mis-registrations occurred in cases where a large gas pocket was present. These gas pockets caused not only significant intrafraction motion of the seminal vesicles, but also led to local signal intensity artefacts which partly overlapped the seminal vesicles. This effect hindered the registration algorithm and led to inaccurate determination of the seminal vesicle location in these dynamics. However, in all cases the gas pocket passed after a few (range 1-5) dynamics (< 1 min) after which the registration algorithm was able to continue registration. Banding artefacts may occur in bTFE sequences due to presence of rectal gas, however in this study we did not observe other significant influences of banding artefacts than the previously stated cases. Moreover, a bandwidth of 434 Hz/px (Table S1) was used which was deemed sufficient to suppress the influence of any geometric distortions in the target area. Next, we did not observe significant influences of sudden motion on the acquired contrast or blurring/ghosting artefacts in the cine-MR dynamics. All reportedmis-registrations in this study were flagged by the adopted Kalman filter. We therefore propose that this filter could be used to identify sudden transitions that could lead to erroneous tracking results. In addition, the filter could also be used as a trigger for beam interruption. The presented results are based on rigid registration with seminal vesicle specific masks. While deformation of the seminal vesicles is seen, this deformation mainly occurs near the seminal vesicle tips, away from the prostate corpus. Most deformation was observed in the anterior-posterior and cranial-caudal direction due to variation of bladder and rectal filling, which is similar to a previous study on seminal vesicle deformations assessed on repeat computed tomography (CT) by van der Wielen et al.23 In their study, deformation of the prostate and seminal vesicles relative to intraprostatic fiducial markers was determined based on CT. They observed the largest seminal vesicle shape variations on the posterior side of the seminal vesicles, followed by shape variations of the seminal vesicle tips. Similar findings were noted by Stenmark et al.25, who concluded that the seminal vesicles exhibit greater variation with increasing distance from the prostate. These conclusions coincide with our findings on seminal vesicle rotations. Due to the fact that the seminal vesicles are connected to the prostate, largest seminal vesicle displacements were found towards the seminal vesicle tips. However, reduced accuracy in tracking seminal vesicle tips due to deformation would not necessarily pose a problem. Kestin et al.9 performed a pathologic analysis to determine the length of tumour involvement in the seminal vesicles. They found a median length of seminal vesicle involvement of 1.0 cm, while in 90% of the positive cases this was limited to the proximal 2.0 cm. In addition, van der Burgt et al.26 studied the impact of tumour invasion on seminal vesicle mobility. They reported that increasing tumour invasion in the seminal vesicles reduces their mobility relative to the prostate corpus. Moreover, van der Burgt et al.26 suggested that the flexibility
RkJQdWJsaXNoZXIy MTk4NDMw