Related Experiment Video
Updated: Jun 22, 2025
Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
Published on: April 11, 2025
Efficient pyramid channel attention network for pathological myopia recognition with pretraining-and-finetuning
Xiaoqing Zhang1, Jilu Zhao2, Yan Li3
1Research Institute of Trustworthy Autonomous Systems and Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen, 518055, China; Center for High Performance Computing and Shenzhen Key Laboratory of Intelligent Bioinformatics, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, 518055, China.
Abstract:
Pathological myopia (PM) is the leading ocular disease for impaired vision worldwide. Clinically, the characteristics of pathology distribution in PM are global-local on the fundus image, which plays a significant role in assisting clinicians in diagnosing PM. However, most existing deep neural networks focused on designing complex architectures but rarely explored the pathology distribution prior of PM. To tackle this issue, we propose an efficient pyramid channel attention (EPCA) module, which fully leverages the potential of the clinical pathology prior of PM with pyramid pooling and multi-scale context fusion. Then, we construct EPCA-Net for automatic PM recognition based on fundus images by stacking a sequence of EPCA modules. Moreover, motivated by the recent pretraining-and-finetuning paradigm, we attempt to adapt pre-trained natural image models for PM recognition by freezing them and treating the EPCA and other attention modules as adapters. In addition, we construct a PM recognition benchmark termed PM-fundus by collecting fundus images of PM from publicly available datasets. The comprehensive experiments demonstrate the superiority of EPCA-Net over state-of-the-art methods in the PM recognition task. For example, EPCA-Net achieves 97.56% accuracy and outperforms ViT by 2.85% accuracy on the PM-fundus dataset. The results also show that our method based on the pretraining-and-finetuning paradigm achieves competitive performance through comparisons to part of previous methods based on traditional fine-tuning paradigm with fewer tunable parameters, which has the potential to leverage more natural image foundation models to address the PM recognition task in limited medical data regime.
Insights
This study introduces EPCA-Net, an efficient deep learning model for recognizing pathological myopia (PM) using fundus images. EPCA-Net leverages clinical pathology distribution priors to achieve high accuracy in diagnosing this leading cause of vision impairment.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Pathological myopia (PM) is a primary cause of vision impairment globally.
- Current deep learning models for PM often overlook crucial pathology distribution priors found in fundus images.
- Effective PM diagnosis relies on understanding global-local pathology patterns.
Purpose of the Study:
- To develop an efficient deep learning model that incorporates clinical pathology distribution priors for pathological myopia recognition.
- To introduce the Efficient Pyramid Channel Attention (EPCA) module and EPCA-Net for enhanced PM detection.
- To evaluate the efficacy of a pretraining-and-finetuning paradigm for PM recognition using limited medical data.
Main Methods:
- Proposed an Efficient Pyramid Channel Attention (EPCA) module integrating pyramid pooling and multi-scale context fusion.
- Constructed EPCA-Net by stacking EPCA modules for automatic PM recognition from fundus images.
- Adapted pre-trained natural image models using EPCA as adapters within a pretraining-and-finetuning framework.
- Created the PM-fundus benchmark dataset for evaluating PM recognition models.
Main Results:
- EPCA-Net achieved 97.56% accuracy on the PM-fundus dataset, outperforming existing state-of-the-art methods.
- The pretraining-and-finetuning approach with EPCA demonstrated competitive performance with fewer tunable parameters.
- The proposed method shows potential for leveraging large natural image models in data-limited medical scenarios.
Conclusions:
- EPCA-Net effectively utilizes pathological myopia distribution priors, offering superior performance in automated fundus image-based recognition.
- The pretraining-and-finetuning strategy with attention modules presents a promising direction for medical image analysis with limited datasets.
- This work provides a robust and efficient tool for pathological myopia diagnosis, aiding in the management of vision impairment.
