Related Experiment Video
Updated: Sep 13, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
RIS-UNet: A Multi-Level Hierarchical Framework for Liver Tumor Segmentation in CT Images
Yuchai Wan1, Lili Zhang1, Murong Wang2
1Beijing Key Laboratory of Big Data Technology for Food Safety, Beijing Technology and Business University, Beijing 100048, China.
Entropy (Basel, Switzerland)
|July 29, 2025
Summary
This study introduces a new deep learning framework to improve liver tumor segmentation accuracy on CT scans. The method enhances diagnostic decision-making by effectively analyzing both inter-slice and inner-slice features.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer-Aided Diagnosis
Background:
- Deep learning analysis of liver CT images aids clinical diagnosis.
- Current methods lack the accuracy needed for clinical requirements.
- Improved liver tumor segmentation is crucial for patient care.
Purpose of the Study:
- To propose a novel multi-level hierarchical framework for enhanced liver tumor segmentation.
- To address the accuracy-efficiency trade-off in existing segmentation strategies.
- To improve the feature representation for more accurate tumor identification.
Main Methods:
- A 2.5D network integrates inter-slice spatial information to balance accuracy and efficiency.
- A Res-Inception-SE Block extracts comprehensive global and local features within slices.
- A hybrid loss function (Binary Cross Entropy and Dice loss) addresses category imbalance and accelerates convergence.
Main Results:
- The proposed framework demonstrates significant improvements in accuracy for liver tumor segmentation.
- Experiments show enhanced efficiency compared to conventional 2D/3D methods.
- The method yields superior visual results on the LiTS17 dataset.
Conclusions:
- The novel multi-level hierarchical framework effectively improves liver tumor segmentation.
- The approach offers a promising solution for clinical decision support in liver imaging.
- This work advances the state-of-the-art in deep learning for medical image analysis.
