This is an early access version, the complete PDF, HTML, and XML versions will be available soon.
Cone-Beam CT Segmentation for Intraoperative Electron Radiotherapy Based on U-Net Variants with Transformer and Extended LSTM Approaches
by
, , ,
Sara Vockner
1,*,
Matthias Mattke
1,
Ivan M. Messner
1,
Christoph Gaisberger
1,2,
Franz Zehentmayr
1
,
Klarissa Ellmauer
1,
Elvis Ruznic
1,
Josef Karner
1,
Gerd Fastner
1,
Roland Reitsamer
3,
Falk Roeder
1,2 and 
Markus Stana
1 1
Department of Radiation Therapy and Radiation Oncology, Paracelsus Medical University, 5020 Salzburg, Austria
2
Institute of Research and Development of Advanced Radiation Technologies (radART), Paracelsus Medical University, 5020 Salzburg, Austria
3
Department of Gynecology, Paracelsus Medical University, 5020 Salzburg, Austria
*
Author to whom correspondence should be addressed.
Cancers 2025, 17(3), 485; https://doi.org/10.3390/cancers17030485
Submission received: 9 December 2024
/
Revised: 23 January 2025
/
Accepted: 27 January 2025
/
Published: 1 February 2025
(This article belongs to the Special Issue Recent Advances in Deep Learning and Medical Imaging for Cancer Treatment (Volume II))
Simple Summary
This study explores the use of advanced neural network architectures to automatically segment CBCT images acquired during intraoperative electron radiotherapy (IOERT). By integrating self-attention and xLSTM features with the U-Net architecture, improved segmentation accuracy for relevant anatomical structures was achieved. These results pave the way for the optimisation of IOERT workflows and facilitate the generation of synthetic CT images. This is a basis for adaptive 3D treatment planning, ultimately enhancing treatment precision in IOERT.
Abstract
AI applications are increasingly prevalent in radiotherapy, including commercial software solutions for automatic segmentation of anatomical structures for 3D 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.
