MSA-MaxNet: A New Model for Medical Image Segmentation
Author Information
Author(s): Wei Wu, Junfeng Huang, Mingxuan Zhang, Yichen Li, Qijia Yu, Qi Zhao
Primary Institution: University of Science and Technology Liaoning
Hypothesis
Can the MSA-MaxNet model improve medical image segmentation by effectively capturing both local and global features?
Conclusion
MSA-MaxNet achieves state-of-the-art performance in medical image segmentation across multiple datasets.
Supporting Evidence
- MSA-MaxNet achieved a Dice score of 85.95% on the Synapse dataset.
- MSA-MaxNet outperformed the second-best model EMCAD by 0.24% on the ACDC dataset.
- MSA-MaxNet achieved a Dice score of 95.26% on the Kvasir-SEG dataset.
Takeaway
This study created a new model called MSA-MaxNet that helps computers better understand medical images, making it easier to find problems like tumors.
Methodology
The study used an encoder-decoder structure with MaxViT blocks and attention mechanisms to enhance feature extraction in medical images.
Limitations
The model has high computational complexity and slower inference speeds compared to CNN-based networks.
Participant Demographics
The study utilized three publicly available medical imaging datasets: Synapse, ACDC, and Kvasir-SEG.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website