A commercial deep learning (DL)-based automated segmentation tool (AST) for computed tomography (CT) is evaluated for accuracy and efficiency gain within prostate cancer patients. Thirty patients from six clinics were reviewed with manual- (MC), automated- (AC) and automated and edited (AEC) contouring methods. In the AEC group, created contours (prostate, seminal vesicles, bladder, rectum, femoral heads and penile bulb) were edited, whereas the MC group included empty datasets for MC. In one clinic, lymph node CTV delineations were evaluated for interobserver variability. Compared to MC, the mean time saved using the AST was 12 min for the whole data set (46%) and 12 min for the lymph node CTV (60%), respectively. The delineation consistency between MC and AEC groups according to the Dice similarity coefficient (DSC) improved from 0.78 to 0.94 for the whole data set and from 0.76 to 0.91 for the lymph nodes. The mean DSCs between MC and AC for all six clinics were 0.82 for prostate, 0.72 for seminal vesicles, 0.93 for bladder, 0.84 for rectum, 0.69 for femoral heads and 0.51 for penile bulb. This study proves that using a general DL-based AST for CT images saves time and improves consistency.

During radiotherapy, prostate motion changes over time. Quantifying and accounting for this motion is essential. This study aimed to assess intra-fraction prostate motion and derive duration-dependent planning margins for two treatment techniques. A four-dimension (4D) transperineal ultrasound Clarity® system was used to track prostate motion. We analysed 1913 fractions from 60 patients undergoing volumetric-modulated arc therapy (VMAT) to the prostate. The mean VMAT treatment duration was 3.4 min. Extended monitoring was conducted weekly to simulate motion during intensity-modulated radiation therapy (IMRT) treatment (an additional seven minutes). A motion-time trend analysis was conducted and the mean intra-fraction motion between VMAT and IMRT treatments compared. Duration-dependent margins were calculated and anisotropic margins for VMAT and IMRT treatments were derived. There were statistically significant differences in the mean intra-fraction motion between VMAT and the simulated IMRT duration in the inferior (0.1 mm versus 0.3 mm) and posterior (−0.2 versus −0.4 mm) directions respectively (p ≪ 0.01). An intra-fraction motion trend inferiorly and posteriorly was observed. The recommended minimum anisotropic margins are 1.7 mm/2.7 mm (superior/inferior); 0.8 mm (left/right), 1.7 mm/2.9 mm (anterior/posterior) for VMAT treatments and 2.9 mm/4.3 mm (superior/inferior), 1.5 mm (left/right), 2.8 mm/4.8 mm (anterior/posterior) for IMRT treatments. Smaller anisotropic margins were required for VMAT compared to IMRT (differences ranging from 1.2 to 1.6 mm superiorly/inferiorly, 0.7 mm laterally and 1.1–1.9 mm anteriorly/posteriorly). VMAT treatment is preferred over IMRT as prostate motion increases with time. Larger margins should be employed in the inferior and posterior directions for both treatment durations. Duration-dependent margins should be applied in the presence of prolonged imaging and verification time.

Inconsistent bladder and rectal volumes have been associated with motion uncertainties during prostate radiotherapy. This study investigates the impact of these volumes to determine the optimal bladder volume. 60 patients from two Asian hospitals were recruited prospectively. 1887 daily cone-beam computed tomography (CBCT) images were analysed. Intra-fraction motion of the prostate was monitored real-time using a four-dimension transperineal ultrasound (4D TPUS) Clarity® system. The impact of planned bladder volume, adequacy of daily bladder filling, and rectum volume on mean intra-fraction motion of the prostate was analysed. Patients' ability to comply with the full bladder hydration protocol and level of frustration was assessed using a questionaire. Acute side effects were assessed using the Common Terminology Criteria for Adverse Events (CTCAE) version 3.0 and quality of life (QoL) assessed using the International Prostate Symptom Score (IPSS). The mean (SD) bladder and rectum volumes achieved during daily treatment were 139.7 cm³ (82.4 cm³) and 53.3 cm³ (18 cm³) respectively. Mean (SD) percentage change from planned CT volumes in bladder volume was reduced by 8.2% (48.7%) and rectum volume was increased by 12.4% (42.2%). Linear Mixed effect model analysis revealed a reduction in intra-fraction motion in both the Sup/Inf (p = 0.008) and Ant/Post (p = 0.0001) directions when the daily bladder was filled between 82 and 113% (3rd Quartiles) of the planned CT volumes. A reduction in intra-fraction motion of the prostate in the Ant/Post direction (z-plane) (p = 0.03) was observed when the planned bladder volume was greater than 200 ml. Patients complied well with the hydration protocol with minimal frustration (mean (SD) scores of 2.1 (1.4) and 1.8 (1.2) respectively). There was a moderate positive correlation (0.496) between mean bladder volume and IPSS reported post-treatment urinary straining (p = 0.001). A planned bladder volume >200 cm³ and daily filling between 82 and 113%, reduced intra-fraction motion of the prostate. The hydration protocol was well tolerated.

Objective . Automatic deformable image registration (DIR) is a critical step in adaptive radiotherapy. Manually delineated organs-at-risk (OARs) contours on planning CT (pCT) scans are deformably registered onto daily cone-beam CT (CBCT) scans for delivered dose accumulation. However, evaluation of registered contours requires human assessment, which is time-consuming and subjects to high inter-observer variability. This work proposes a deep learning model that allows accurate prediction of Dice similarity coefficients (DSC) of registered contours in prostate radiotherapy. Approach . Our dataset comprises 20 prostate cancer patients with 37–39 daily CBCT scans each. The pCT scans and planning contours were deformably registered to each corresponding CBCT scan to generate virtual CT (vCT) scans and registered contours. The DSC score, which is a common contour-based validation metric for registration quality, between the registered and manual contours were computed. A Siamese neural network was trained on the vCT-CBCT image pairs to predict DSC. To assess the performance of the model, the root mean squared error (RMSE) between the actual and predicted DSC were computed. Main results . The model showed promising results for predicting DSC, giving RMSE of 0.070, 0.079 and 0.118 for rectum, prostate, and bladder respectively on the holdout test set. Clinically, a low RMSE implies that the predicted DSC can be reliably used to determine if further DIR assessment from physicians is required. Considering the event where a registered contour is classified as poor if its DSC is below 0.6 and good otherwise, the model achieves an accuracy of 92% for the rectum. A sensitivity of 0.97 suggests that the model can correctly identify 97% of poorly registered contours, allowing manual assessment of DIR to be triggered. Significance . We propose a neural network capable of accurately predicting DSC of deformably registered OAR contours, which can be used to evaluate eligibility for plan adaptation.