Related Experiment Video
Updated: May 21, 2025
04:48
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
335
AttmNet: a hybrid Transformer integrating self-attention, Mamba, and multi-layer convolution for enhanced lesion
Hancan Zhu1,2, Yibing Huang2, Kelin Yao1
1Interdisciplinary Research Center, Affiliated Hospital of Shaoxing University, Shaoxing, China.
Quantitative Imaging in Medicine and Surgery
|May 19, 2025
Summary
AttmNet, a novel segmentation network, enhances medical image analysis by integrating Convolutional Neural Networks (CNNs), Transformers, and Mamba. This approach improves lesion segmentation accuracy for better cancer diagnosis and treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Accurate medical image segmentation is crucial for cancer diagnosis and treatment.
- Convolutional Neural Networks (CNNs) struggle with long-range dependencies.
- Transformers capture long-range dependencies but are computationally expensive; Mamba models long-range dependencies efficiently but lacks fine detail precision.
Purpose of the Study:
- To develop a novel segmentation approach combining CNNs, Transformers, and Mamba.
- To enhance global context understanding and local feature extraction in medical image segmentation.
- To improve the accuracy and efficiency of lesion segmentation in medical imaging.
Main Methods:
- Proposed AttmNet, a U-shaped network for medical image segmentation.
- Introduced the Multiscale-Convolution, Self-Attention, and Mamba (MAM) block.
- Evaluated AttmNet on breast, skin, and lung lesion segmentation datasets.
Main Results:
- AttmNet outperformed state-of-the-art methods in intersection over union (IoU) and Dice similarity coefficients.
- Achieved significant improvements on breast ultrasound (BUS) and breast ultrasound images (BUSI) datasets.
- Demonstrated superior performance on dermoscopy and COVID-19 lung lesion segmentation datasets.
Conclusions:
- AttmNet is an efficient and accurate tool for medical image segmentation.
- The MAM block enhances segmentation accuracy and maintains computational efficiency.
- AttmNet shows high suitability for clinical applications.
