MRI-guided prostate adaptive radiotherapy - A systematic review

Dose escalated radiotherapy improves outcomes for men with prostate cancer. A plateau for benefit from dose escalation using EBRT may not have been reached for some patients with higher risk disease. The use of increasingly conformal techniques, such as step and shoot IMRT or more recently VMAT, has allowed treatment intensification to be achieved whilst minimising associated increases in toxicity to surrounding normal structures. To support further safe dose escalation, the uncertainties in the treatment target position will need be minimised using optimal planning and image-guided radiotherapy (IGRT). In particular the increasing usage of profoundly hypo-fractionated stereotactic therapy is predicated on the ability to confidently direct treatment precisely to the intended target for the duration of each treatment. This article reviews published studies on the influences of varies types of motion on daily prostate position and how these may be mitigated to improve IGRT in future. In particular the role that MRI has played in the generation of data is discussed and the potential role of the MR-Linac in next-generation IGRT is discussed.

Automatic detection system for multiple region of interest registration to account for posture changes in head and neck radiotherapy

Despite immobilization of head and neck (H and N) cancer patients, considerable posture changes occur over the course of radiotherapy (RT). To account for the posture changes, we previously implemented a multiple regions of interest (mROIs) registration system tailored to the H and N region for image-guided RT correction strategies. This paper is focused on the automatic segmentation of the ROIs in the H and N region. We developed a fast and robust automatic detection system suitable for an online image-guided application and quantified its performance. The system was developed to segment nine high contrast structures from the planning CT including cervical vertebrae, mandible, hyoid, manubrium of sternum, larynx and occipital bone. It generates nine 3D rectangular-shaped ROIs and informs the user in case of ambiguities. Two observers evaluated the robustness of the segmentation on 188 H and N cancer patients. Bland-Altman analysis was applied to a sub-group of 50 patients to compare the registration results using only the automatically generated ROIs and those manually set by two independent experts. Finally the time performance and workload were evaluated. Automatic detection of individual anatomical ROIs had a success rate of 97%/53% with/without user notifications respectively. Following the notifications, for 38% of the patients one or more structures were manually adjusted. The processing time was on average 5 s. The limits of agreement between the local registrations of manually and automatically set ROIs was comprised between ±1.4 mm, except for the manubrium of sternum (-1.71 mm and 1.67 mm), and were similar to the limits agreement between the two experts. The workload to place the nine ROIs was reduced from 141 s (±20 s) by the manual procedure to 59 s (±17 s) using the automatic method. An efficient detection system to segment multiple ROIs was developed for Cone-Beam CT image-guided applications in the H and N region and is clinically implemented in our department.

Validation of deformable registration in head and neck cancer using analysis of variance

PURPOSE

Deformable image registration (DIR) is often validated based on a distance-to-agreement (DTA) criterion of automatically propagated anatomical landmarks that were manually identified. Due to human observer variability, however, the performance of the registration method is diluted. The purpose of this study was to evaluate an analysis of variance (ANOVA) based validation to account for such observer variation.

METHODS

Weekly cone beam CTs (CBCTs) of ten head and neck cancer patients undergoing five weeks of radiotherapy were used. An expert identified 23 anatomical features (landmarks) on the planning CT. The landmarks were automatically propagated to the CBCT using multiregion-of-interest (mROI) registration. Additionally, two human observers independently localized these landmarks on the CBCTs. Subsequently, ANOVA was used to compute the variance of each observer on the pairwise distance (PWD).

RESULTS

ANOVA based analysis demonstrated that a classical DTA approach underestimated the precision for the mROI due to human observer variation by about 25%. The systematic error (accuracy) of mROI ranged from 0.13 to 0.17 mm; the variability (1 SD) (precision) ranged from 1.3 to 1.5 mm demonstrating that its performance is dominated by the precision.

CONCLUSIONS

The PWD-ANOVA method accounts for human observer variation allowing a better estimation of the of DIR errors.