Related Experiment Video
Updated: Nov 16, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.1K
Contour-aware semantic segmentation network with spatial attention mechanism for medical image
Zhiming Cheng1, Aiping Qu1,2, Xiaofeng He1
1School of Computer, University of South China, Hengyang, 421001 China.
The Visual Computer
|March 1, 2021
Summary
This study introduces a novel contour-aware semantic segmentation network, an extension of Unet, for improved medical image segmentation. The enhanced network accurately extracts semantic features and refines contour details, outperforming existing methods.
Area of Science:
- Medical Imaging
- Computer-Aided Diagnosis
- Deep Learning
Background:
- Medical image segmentation is crucial for computer-aided systems but remains challenging due to diverse modalities and cases.
- Unet is a popular deep learning framework for accurate biomedical image segmentation.
Purpose of the Study:
- To propose a novel contour-aware semantic segmentation network for enhanced medical image segmentation.
- To improve the accuracy and detail extraction in medical image segmentation tasks.
Main Methods:
- The proposed network extends Unet with a semantic branch for feature extraction and a detail branch for contour enhancement.
- A MulBlock module was designed for extracting semantic information with varied receptive fields.
- A Channel Attention Module (CAM) was employed to adaptively suppress redundant features.
Main Results:
- The contour-aware network demonstrated remarkable performance on public medical image segmentation challenges.
- The method effectively extracts semantic features and enhances contour details compared to state-of-the-art approaches.
Conclusions:
- The proposed contour-aware semantic segmentation network offers a significant advancement in medical image segmentation.
- This approach shows promise for improving the development of clinical computer-aided diagnostic systems.
