Related Experiment Video
Updated: Aug 3, 2026
06:48
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
8.8K
Comprehensive Segmentation of Gray Matter Structures on T1-Weighted Brain MRI: A Comparative Study of Convolutional
Yujia Wei1, Jaidip Manikrao Jagtap1, Yashbir Singh1
1From the Department of Radiology, Mayo Clinic, Rochester, Minnesota.
AJNR. American Journal of Neuroradiology
|October 21, 2024
Summary
U-Mamba_Bot excels at segmenting 122 brain structures from MRI scans, outperforming other deep learning models. This advanced AI tool shows promise for understanding neurodegenerative diseases like Alzheimer's.
Area of Science:
- Artificial Intelligence
- Neuroimaging
- Medical Image Analysis
Background:
- Deep learning models show promise in medical image segmentation.
- Existing brain MRI segmentation models have limitations in dataset size and structure identification.
- This study aims to identify the most effective deep learning model for segmenting numerous brain structures.
Purpose of the Study:
- To evaluate and compare the performance of six advanced deep learning models for segmenting 122 brain structures from T1-weighted MRI scans.
- To determine the most effective model for clinical and research applications in brain MRI segmentation.
- To assess the potential of these models in understanding neurodegenerative diseases.
Main Methods:
- 1510 T1-weighted MRI scans were analyzed.
- Six deep learning models (nnU-Net, SegResNet, SwinUNETR, UNETR, U-Mamba_BOT, U-Mamba_Enc) were compared.
- Accuracy was measured using Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD95).
- Brain structure volumes were compared between normal controls and Alzheimer's disease patients.
Main Results:
- U-Mamba_Bot achieved the highest median DSC (0.9112).
- nnU-Net demonstrated the highest HD95 (1.392), indicating precise structure capture.
- Volume analysis revealed changes consistent with existing literature for Alzheimer's disease patients.
Conclusions:
- U-Mamba_Bot is a highly effective tool for detailed brain structure segmentation in T1-weighted MRI.
- Deep learning models show potential for advancing the understanding of neurodegenerative diseases like Alzheimer's.
- Further validation with larger datasets and exploration in other neurological conditions are recommended.
