Related Experiment Video
Updated: Sep 25, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Artificial Intelligence Radiotherapy Planning: Automatic Segmentation of Human Organs in CT Images Based on a
Guosheng Shen1,2,3,4, Xiaodong Jin1,2,3,4, Chao Sun1,2,3,4
1Institute of Modern Physics, Chinese Academy of Sciences, Lanzhou, China.
Frontiers in Public Health
|May 2, 2022
Summary
A novel deep learning algorithm, BCDU-Net, accurately segments 17 human organ types for radiation therapy planning. This automated organ segmentation significantly speeds up treatment design, meeting clinical requirements.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiotherapy
Background:
- Accurate segmentation of organs at risk (OARs) is crucial for effective radiation therapy planning.
- Current manual segmentation is time-consuming and prone to inter-observer variability.
Purpose of the Study:
- To investigate an automatic organ segmentation technique using a deep learning convolutional neural network (CNN).
- To develop and validate a modified BCDU-Net algorithm for rapid and precise radiotherapy treatment planning.
Main Methods:
- A modified BCDU-Net CNN algorithm was developed.
- The algorithm was trained and validated using 22,000 CT images and manual organ contours from 329 patients.
- Performance was tested on a separate set of CT images.
Main Results:
- The modified BCDU-Net achieved an average Dice similarity coefficient (DSC) of 0.8376 across 17 organ types.
- The highest DSC reached 0.9676.
- Automatic segmentation of 17 organs took approximately 1 hour, significantly faster than manual methods.
Conclusions:
- The modified deep neural network algorithm enables fast and accurate automatic segmentation of 17 human organ types.
- The developed method meets the accuracy and speed requirements for radiotherapy applications.
