Target volume delineation is a vital but time-consuming and challenging part of radiotherapy, where the goal is to deliver sufficient dose to the target while reducing risks of side effects. For head and neck cancer (HNC) this is complicated by the complex anatomy of the head and neck region and the proximity of target volumes to organs at risk. The purpose of this study was to compare and evaluate conventional PET thresholding methods, six classical machine learning algorithms and a 2D U-Net convolutional neural network (CNN) for automatic gross tumor volume (GTV) segmentation of HNC in PET/CT images. For the latter two approaches the impact of single versus multimodality input on segmentation quality was also assessed. 197 patients were included in the study. The cohort was split into training and test sets (157 and 40 patients, respectively). Five-fold cross-validation was used on the training set for model comparison and selection. Manual GTV delineations represented the ground truth. Tresholding, classical machine learning and CNN segmentation models were ranked separately according to the cross-validation Sørensen–Dice similarity coefficient (Dice). PET thresholding gave a maximum mean Dice of 0.62, whereas classical machine learning resulted in maximum mean Dice scores of 0.24 (CT) and 0.66 (PET; PET/CT). CNN models obtained maximum mean Dice scores of 0.66 (CT), 0.68 (PET) and 0.74 (PET/CT). The difference in cross-validation Dice between multimodality PET/CT and single modality CNN models was significant (p ≤ 0.0001). The top-ranked PET/CT-based CNN model outperformed the best-performing thresholding and classical machine learning models, giving significantly better segmentations in terms of cross-validation and test set Dice, true positive rate, positive predictive value and surface distance-based metrics (p ≤ 0.0001). Thus, deep learning based on multimodality PET/CT input resulted in superior target coverage and less inclusion of surrounding normal tissue.

background. Dose painting by numbers (DPBN) is a method to deliver an inhomogeneous tumor dose voxel-by-voxel with a prescription based on biological medical images. However, planning of DPBN is not supported by commercial treatment planning systems (TPS) today. Here, a straightforward method for DPBN with a standard TPS is presented. Material and methods. DPBN tumor dose prescription maps were generated from 18 F-FDG-PET images applying a linear relationship between image voxel value and dose. An inverted DPBN prescription map was created and imported into a standard TPS where it was defined as a mock pre-treated dose. Using inverse optimization for the summed dose, a planned DPBN dose distribution was created. The procedure was tested in standard TPS for three different tumor cases; cervix, lung and head and neck. The treatment plans were compared to the prescribed DPBN dose distribution by three-dimensional (3D) gamma analysis and quality factors (QFs). Delivery of the DPBN plans was assessed with portal dosimetry (PD). results. Maximum tumor doses of 149%, 140% and 151% relative to the minimum tumor dose were prescribed for the cervix, lung and head and neck case, respectively. DPBN distributions were well achieved within the tumor whilst normal tissue doses were within constraints. Generally, high gamma pass rates ( 89% at 2%/2 mm) and low QFs ( 2.6%) were found. PD showed that all DPBN plans could be successfully delivered. conclusions. The presented methodology enables the use of currently available TPSs for DPBN planning and delivery and may therefore pave the way for clinical implementation.

No abstract