Related Experiment Video
Updated: Sep 5, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
CAFS: An Attention-Based Co-Segmentation Semi-Supervised Method for Nasopharyngeal Carcinoma Segmentation
Yitong Chen1, Guanghui Han1,2, Tianyu Lin1
1School of Biomedical Engineering, Sun Yat-sen University, Shenzhen 518107, China.
Sensors (Basel, Switzerland)
|July 9, 2022
Summary
This study introduces CAFS, a semi-supervised method for segmenting nasopharyngeal carcinoma (NPC) tumors. CAFS effectively addresses data scarcity and tumor similarity challenges, achieving superior accuracy in segmenting nasopharyngeal carcinoma.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Accurate nasopharyngeal carcinoma segmentation is critical for effective treatment.
- Existing deep learning methods face challenges like limited labeled data, tumor similarity to surrounding tissues, and complex tumor shapes.
Purpose of the Study:
- To propose a novel semi-supervised method, CAFS, for automatic nasopharyngeal carcinoma segmentation.
- To overcome the limitations of current segmentation techniques, particularly data scarcity.
Main Methods:
- Developed a semi-supervised method named CAFS.
- Incorporated a teacher-student cooperative segmentation mechanism, an attention mechanism, and a feedback mechanism.
- Utilized a small amount of labeled nasopharyngeal carcinoma data.
Main Results:
- CAFS achieved an average Dice Similarity Coefficient (DSC) of 0.8723 for nasopharyngeal carcinoma segmentation.
- Outperformed state-of-the-art methods in segmentation accuracy.
- Demonstrated superior DSC, Jaccard, and precision values compared to existing methods, with DSC 7.42% higher than the best previous results.
Conclusions:
- CAFS effectively segments nasopharyngeal carcinoma even with limited labeled data.
- The proposed method shows significant improvements over current state-of-the-art techniques.
- CAFS offers a promising solution for accurate nasopharyngeal carcinoma segmentation in clinical settings.
