Reproducibility of the bladder shape and bladder shape changes during filling

Abstract

The feasibility of high precision radiotherapy to the bladder region is limited by bladder motion and volume changes. In the near future, we plan to begin treatment delivery of bladder cancer patients with the acquisition of a cone beam CT image on which the complete bladder will be semi-automatically localized. Subsequently, a bladder shape model that was developed in a previous study will be used for bladder localization and for the prediction of shape changes in the time interval between acquisition and beam delivery. For such predictions, knowledge about urinary inflow rate is required. Therefore, a series of MR images was acquired over 1 h with time intervals of 10 min for 18 healthy volunteers. To gain insight in the reproducibility of the bladder shape over longer periods of time, two additional MRI series were recorded for 10 of the volunteers. To a good approximation, the bladder volume increased linearly in time for all individuals. Despite receiving drinking instructions, we found a large variation in the inflow rate between individuals, ranging from 2.1 to 15 cc/min (mean value: 9 +/- 3 cc/min). In contrast, the intravolunteer variation was much smaller, with a mean standard deviation (SD) of 0.4 cc/min. The inflow rate was linearly correlated with age (negative slope). To study the reproducibility of the bladder shape, we compared bladder shapes of equal volume. For all individuals, the caudal part of the bladder was the most reproducible (variations<0.3 cm in all cases). The cranial and posterior parts of the bladder was much less reproducible, with local SD values up to approximately 1.2 cm for bladders with a volume of 200 cc. These large long-term variations were primarily caused by changes in position and filling of the small bowel and rectum. However, for short time intervals, the rectal filling was (nearly) constant. Therefore, the reproducibility of urinary inflow, combined with the previously developed shape model gives us an excellent tool to predict short-term shape changes. We intend to use this tool for further improvement of image-guided radiotherapy for bladder cancer patients.