Related Experiment Video
Updated: Aug 21, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
MC-UNet: Multimodule Concatenation Based on U-Shape Network for Retinal Blood Vessels Segmentation
Jun Li1,2, Ting Zhang1,2, Yi Zhao1
1College of Computer and Information, Fujian Agriculture and Forestry University, Fuzhou 350002, China.
Computational Intelligence and Neuroscience
|November 17, 2022
Summary
This study introduces a novel U-Net based deep learning model for retinal blood vessel segmentation. The method enhances accuracy, particularly for microvessels, aiding ophthalmic disease diagnosis.
Area of Science:
- Medical Imaging
- Computer Vision
- Ophthalmology
Background:
- Accurate retinal blood vessel segmentation is crucial for diagnosing ophthalmic diseases.
- Existing deep learning methods face challenges due to complex vascular structures and ambiguous pathological features.
Purpose of the Study:
- To propose a novel multimodule concatenation U-Net for improved retinal blood vessel segmentation.
- To enhance the capture of contextual information and expand the receptive field for more precise segmentation.
Main Methods:
- A U-shaped network incorporating atrous convolution and multikernel pooling.
- A multimodule concatenation integrating spatial attention, dense atrous convolution, and multikernel pooling blocks.
- Cascading atrous convolutions with varying dilation rates to increase the receptive field.
Main Results:
- The proposed method demonstrates effectiveness in retinal blood vessel segmentation on public datasets (DRIVE, STARE, CHASE_DB1).
- Significant improvements were observed, especially in segmenting delicate microvessels.
- The network successfully captures more contextual information and larger receptive fields.
Conclusions:
- The novel multimodule concatenation U-Net is effective for retinal blood vessel segmentation.
- The approach shows particular promise for accurate microvessel segmentation in clinical applications.
- The developed method advances automated analysis in ophthalmology.
