Related Experiment Video
Updated: May 21, 2025
06:53
3D Whole-heart Myocardial Tissue Analysis
Published on: April 12, 2017
8.7K
An enhanced visual state space model for myocardial pathology segmentation in multi-sequence cardiac MRI
Shuning Li1,2,3, Xiang Li1,2,3, Pingping Wang1,2,3
1Center for Medical Artificial Intelligence, Shandong University of Traditional Chinese Medicine, Qingdao, China.
Medical Physics
|March 20, 2025
Summary
A new deep learning model, MPS-Mamba, enhances myocardial pathology segmentation in cardiac MRI scans. This novel approach improves accuracy for scar and edema detection, offering a potential tool for clinical diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiology
Background:
- Myocardial pathology segmentation is vital for diagnosing and treating myocardial infarction (MI).
- Current models struggle with cardiac magnetic resonance (CMR) images due to scale variations and multiple objects.
- Limitations include insufficient remote modeling in CNNs and high computational cost in transformers.
Purpose of the Study:
- To develop a novel model for accurate and efficient myocardial pathology segmentation in CMR images.
- To address challenges posed by image characteristics and algorithmic limitations in current methods.
Main Methods:
- Developed MPS-Mamba, a visual state space (VSS)-based deep neural network.
- Employed a dual-branch encoder (VSS and convolutional) for global and local feature extraction.
- Integrated multi-scale feature fusion and a decoder with constraint functions for anatomical accuracy.
Main Results:
- MPS-Mamba achieved superior performance on the MyoPS 2020 dataset.
- Demonstrated high Dice scores for myocardial scar (0.717 ± 0.169) and edema (0.735 ± 0.073) segmentation.
- Validated effectiveness in multi-sequence CMR images, outperforming mainstream methods.
Conclusions:
- MPS-Mamba shows significant effectiveness and superiority in myocardial pathology segmentation.
- The method is a promising tool for assisting clinical diagnosis of myocardial infarction.
- Potential to improve accuracy and efficiency in clinical workflows.
