Artificial Intelligence (AI) applications are increasingly prevalent in radiotherapy, including commercial software solutions for automatic segmentation of anatomical structures for 3D Computed Tomography (CT). However, their use in intraoperative electron radiotherapy (IOERT) remains limited. In particular, no AI solution is available for contouring cone beam CT (CBCT) images acquired with a mobile CBCT device. The U-Net convolutional neural network architecture has gained huge success for medical image segmentation but still has difficulties capturing the global context. To increase the accuracy in CBCT segmentation for IOERT, three different AI architectures were trained and evaluated. The features of the natural language processing models Transformer and xLSTM were added to the popular U-Net architecture and compared with the standard U-Net and manual segmentation performance. These networks were trained and tested using 55 CBCT scans obtained from breast cancer patients undergoing IOERT in the department of radiotherapy and radiation oncology in Salzburg, and each architecture’s segmentation performance was assessed using the dice coefficient (DSC) as a similarity measure. The average DSC values were 0.83 for the standard U-Net, 0.88 for the U-Net with transformer features, and 0.66 for the U-Net with xLSTM. The hybrid U-Net architecture, including Transformer features, achieved the best segmentation accuracy, demonstrating an improvement of 5% on average over the standard U-Net, while the U-Net with xLSTM showed inferior performance compared to the standard U-Net. With the help of automatic contouring, synthetic CT images can be generated, and IOERT challenges related to the time-consuming nature of 3D image-based treatment planning can be addressed.
Background/Objectives: Accurate characterization of treatment geometry is essential for intraoperative electron radiotherapy (IOeRT) treatment planning. Mobile cone-beam computed tomography (CBCT) enables volumetric assessment of intraoperative anatomy beyond conventional point-based measurements. This study compared clinically applied distance measurement techniques with geometric distance descriptors retrospectively extracted from comprehensive three-dimensional CBCT datasets. Methods: Ninety-six patients receiving IOeRT as a tumor-bed boost during breast-conserving surgery were included. Manual CBCT measurements (mCBCT) were compared with mean (USavg) and minimum (USmin) ultrasound measurements per patient. Comprehensive 3D distance extraction was performed in 82 CBCT datasets using an approximated thoracic-wall surface as the geometric reference. Minimum and first-quartile distances were evaluated across the complete applicator cross-section and within a 1-cm-diameter inner disk around the central beam axis. Results: The mean paired difference between mCBCT and USavg was 0.3 mm and was not statistically significant, whereas mCBCT measurements exceeded USmin by an average of 1.5 mm. In the principal comparison, the minimum CBCT-derived distance within the inner disk was 1.4 mm smaller than mCBCT. Across the complete applicator area, minimum distances were 9.1-10.7 mm smaller than the clinical measurements. The first-quartile distance across all surface points was 1.4 mm smaller than mCBCT, whereas the corresponding inner-disk difference was not statistically significant after correction. Isodose-atlas-based dose-volume analysis indicated that a 1.5-mm change in assumed penetration depth could correspond to an approximately 7% change in the 90% isodose volume at shallow depths. Conclusions: Clinical CBCT and ultrasound measurements showed small group-level differences but substantial patient-level variability. Comprehensive 3D CBCT analysis revealed spatial variations in tissue depth not fully represented by the clinical measurements, highlighting the limitations of characterizing treatment geometry using a single distance. Although anatomical accuracy and patient-specific dosimetric consequences remain to be established, these findings support the potential of intraoperative 3D CBCT to advance IOeRT toward individualized, image-guided treatment planning.