Related Experiment Video
Updated: Jun 17, 2025
08:34
Proton Therapy Delivery and Its Clinical Application in Select Solid Tumor Malignancies
Published on: February 6, 2019
20.3K
Deep learning based clinical target volumes contouring for prostate cancer: Easy and efficient application
Feng Wen1,2, Zhebin Chen3,4, Xin Wang1,2
1Department of Radiation Oncology, Cancer Center, West China Hospital, Sichuan University, Chengdu, China.
Journal of Applied Clinical Medical Physics
|August 9, 2024
Summary
This study introduces an AI model for automated prostate cancer radiotherapy contouring, significantly reducing delineation time and improving consistency compared to manual methods. The deep learning approach enhances efficiency and accuracy in defining clinical target volumes (CTVs) for both radical and postoperative treatments.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Oncology
- Radiation Therapy Planning
Background:
- Manual segmentation of clinical target volumes (CTVs) in prostate cancer radiotherapy is time-consuming and prone to inter-observer variability.
- Accurate delineation of CTVs is critical for effective radiation delivery and minimizing side effects.
Purpose of the Study:
- To develop and evaluate a deep learning-based automated contouring model for CTVs in prostate cancer radiotherapy.
- To compare the performance of the AI model against manual delineations by radiation oncologists.
Main Methods:
- A deep learning model incorporating an attention mechanism was developed using computed tomography (CT) data from 197 prostate cancer patients.
- Two models were created: one for radical radiotherapy (including CTVn for lymph nodes and CTVp for prostate) and one for postoperative radiotherapy.
- The AI model's accuracy and efficiency were evaluated against junior and senior radiation oncologists' manual delineations.
Main Results:
- The AI model demonstrated superior or comparable accuracy in delineating CTVn and CTVp compared to junior physicians, evidenced by higher volumetric dice coefficients.
- AI significantly reduced delineation time, with median times of 0.23-0.26 minutes versus over 45 minutes for physicians (p < 0.001).
- Senior physician correction time for AI-generated contours was substantially shorter than for manual delineations (p < 0.001).
Conclusions:
- A deep learning and attention mechanism-based automated contouring model provides a highly consistent and time-saving solution for CTV delineation in prostate cancer.
- This AI approach can improve efficiency in radiotherapy planning and may serve as a valuable tool for training junior radiation oncologists.
