Related Experiment Video
Updated: Jun 6, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.7K
RMCNet: A Liver Cancer Segmentation Network Based on 3D Multi-Scale Convolution, Attention, and Residual Path
Zerui Zhang1, Jianyun Gao2, Shu Li3
1School of Bioengineering, Chongqing University, Chongqing 400044, China.
Bioengineering (Basel, Switzerland)
|November 27, 2024
Summary
Researchers developed RMCNet, a 3D segmentation algorithm for liver cancer detection in CT scans. This novel approach improves tumor segmentation accuracy by effectively handling variations in tumor size, shape, and location.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Abdominal CT scans are crucial for diagnosing liver cancer.
- Accurate segmentation of liver tumors is challenging due to variations in size, shape, and location.
- Existing segmentation methods may struggle with these complexities.
Purpose of the Study:
- To propose an end-to-end 3D segmentation algorithm, RMCNet, for improved liver cancer lesion detection.
- To address the challenges of varying tumor characteristics in CT image segmentation.
- To enhance the accuracy and generalization capability of liver tumor segmentation.
Main Methods:
- Developed RMCNet, an end-to-end 3D segmentation algorithm.
- Incorporated a 3D multiscale convolution module for extracting tumors of various sizes.
- Utilized a convolutional block attention module (CBAM) to focus on tumor shape and location.
- Integrated residual paths in encoding layers to enrich feature maps.
Main Results:
- Achieved high Dice Similarity Coefficient (DSC) scores: 76.56% and 72.96%.
- Obtained high Jaccard Coefficient (JCC) scores: 75.82% and 71.25%.
- Demonstrated low Hausdorff Distance (HD) and Average Surface Distance (ASD) values, indicating precise segmentation.
Conclusions:
- RMCNet shows superior performance in segmenting liver tumors compared to existing methods.
- The algorithm exhibits strong generalization capabilities across different datasets.
- RMCNet offers a promising solution for accurate liver cancer lesion segmentation in abdominal CT images.
