Related Experiment Video
Updated: Jun 7, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
MIU-Net: Advanced multi-scale feature extraction and imbalance mitigation for optic disc segmentation.
Yichen Xiao1, Yi Shao1, Zhi Chen1
1Eye Institute and Department of Ophthalmology, Eye and ENT Hospital, Fudan University, Shanghai, 200031, China; NHC Key Laboratory of Myopia (Fudan University), Key Laboratory of Myopia, Chinese Academy of Medical Sciences, Shanghai, 200031, China; Shanghai Research Center of Ophthalmology and Optometry, Shanghai, 200031, China; Shanghai Key Laboratory of Visual Impairment and Restoration, Shanghai, 200031, China.
A new model, MIU-Net, enhances optic disc segmentation for pathological myopia detection. It improves accuracy in identifying early signs of this severe eye condition, aiding clinical diagnosis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Pathological myopia is a severe eye condition with vision-threatening complications.
- Accurate optic disc segmentation is crucial for early detection but challenging due to image complexities.
- Existing segmentation methods lack the required clinical accuracy.
Purpose of the Study:
- To develop an advanced model for accurate optic disc segmentation in pathological myopia.
- To improve the identification of subtle optic disc changes indicative of the disease.
- To enhance diagnostic capabilities for pathological myopia through improved medical image analysis.
Main Methods:
- Proposed MIU-Net model incorporating a multi-scale feature extraction (MFE) module.
- Implemented a dual attention mechanism (channel and spatial) for focused feature utilization.
- Utilized focal loss to address class imbalance and data augmentation for increased data diversity.
Main Results:
- MIU-Net demonstrated significant improvements in segmentation accuracy and robustness.
- The model outperformed existing methods on the iChallenge-PM and iChallenge-AMD datasets.
- Enhanced ability to detect minority optic disc pixels crucial for diagnosis.
Conclusions:
- MIU-Net offers a promising solution for accurate optic disc segmentation in pathological myopia.
- The model shows potential for clinical application in diagnosing eye conditions.
- This approach can advance medical image processing for various diagnostic tasks.
