Related Experiment Video
Updated: Jul 12, 2026
04:48
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
SLIC-Former: a superpixel-guided transformer framework for automatic liver segmentation in CT images.
Sarah F Elqersh1,2, Amira Y Haikal3, Mahmoud M Saafan3
1Department of Computers and Systems, Faculty of Engineering, Mansoura University, Mansoura, 35516, Egypt. sarahelqersh174@std.mans.edu.eg.
Scientific Reports
|July 10, 2026
Summary
This study introduces SLIC-Former, a novel transformer framework for precise liver segmentation in computed tomography (CT) scans. The method enhances accuracy and efficiency by using superpixels, improving anatomical boundary adherence.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Accurate liver segmentation in CT scans is challenging due to irregular organ shape and similar adjacent tissue appearance.
- Existing methods struggle with efficiency and precision in abdominal CT image analysis.
Purpose of the Study:
- To develop an accurate and efficient automatic liver segmentation framework for abdominal CT scans.
- To introduce SLIC-Former, a superpixel-guided transformer model for improved segmentation.
Main Methods:
- Proposed SLIC-Former, a framework utilizing Simple Linear Iterative Clustering (SLIC) for superpixel generation.
- Replaced fixed image patches with adaptive superpixels to align with anatomical boundaries and reduce computation.
- Evaluated on the Liver Tumor Segmentation (LiTS) dataset.
Main Results:
- Achieved a Dice coefficient of 0.93, an Intersection over Union (IoU) of 0.87, and a Volumetric Overlap Error (VOE) of 13.5%.
- Demonstrated high overlap with expert annotations and produced smooth, coherent liver masks.
- SLIC-Former proved computationally efficient compared to traditional patch-based methods.
Conclusions:
- SLIC-Former provides an accurate and efficient tool for automatic liver segmentation in CT images.
- The superpixel-guided approach offers a promising foundation for segmenting other organs and enhancing clinical decision support systems.
