Related Experiment Video
Updated: Aug 1, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.8K
BCU-Net: Bridging ConvNeXt and U-Net for medical image segmentation
Hongbin Zhang1, Xiang Zhong1, Guangli Li2
1School of Software, East China Jiaotong University, China.
Computers in Biology and Medicine
|April 26, 2023
Summary
BCU-Net enhances medical image segmentation by combining ConvNeXt and U-Net for better local and global feature analysis. This novel approach improves diagnostic accuracy, especially for imbalanced datasets.
Area of Science:
- Medical Imaging
- Computer Vision
- Artificial Intelligence
Background:
- Medical image segmentation is crucial for accurate diagnosis.
- U-Net models excel at local feature extraction.
- Limitations exist in exploring global semantics and addressing class imbalance.
Purpose of the Study:
- To introduce BCU-Net, a novel model for medical image segmentation.
- To leverage heterogeneous neural networks for complementary local and global pathological semantics.
- To address the class-imbalance problem in medical image analysis.
Main Methods:
- Developed BCU-Net, integrating ConvNeXt for global interaction and U-Net for local processing.
- Introduced a multilabel recall loss (MRL) module to mitigate class imbalance.
- Conducted extensive experiments on six diverse medical image datasets.
Main Results:
- BCU-Net demonstrated superior performance in qualitative and quantitative evaluations.
- The model showed excellent generalizability across various medical image datasets.
- BCU-Net effectively handles diverse image resolutions and exhibits flexibility.
Conclusions:
- BCU-Net offers a significant advancement in medical image segmentation.
- The model's architecture and MRL module effectively fuse local and global semantics.
- BCU-Net's adaptability and performance highlight its clinical potential.
