Related Experiment Video
Updated: Jun 24, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Deep learning based retinal vessel segmentation and hypertensive retinopathy quantification using heterogeneous
Xinghui Liu1,2, Hongwen Tan2, Wu Wang3
1School of Clinical Medicine, Guizhou Medical University, Guiyang, China.
Insights
This study introduces a novel deep learning model for precise retinal vessel segmentation, crucial for diagnosing diseases like hypertensive retinopathy. The advanced network improves accuracy, especially with lower-quality fundus images.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Retinal vessels are key biomarkers for detecting diseases such as hypertensive retinopathy.
- Manual segmentation of retinal vessels is laborious and time-consuming.
- Image quality significantly impacts the accuracy of automated vessel segmentation.
Purpose of the Study:
- To develop an advanced deep learning model for accurate retinal vessel segmentation.
- To improve the detection and quantification of hypertensive retinopathy.
- To address challenges posed by sub-optimal fundus image quality.
Main Methods:
- Proposed a heterogeneous neural network integrating Convolutional Neural Networks (CNNs) and Transformer networks.
- Utilized a cross-attention mechanism to effectively mine long-range spatial features.
- Employed deep learning methodologies for precise segmentation of retinal vessels.
Main Results:
- The proposed model demonstrated superior performance in vessel segmentation across four public datasets.
- The network effectively handles variations in retinal image quality.
- Significant potential for accurate hypertensive retinopathy quantification was observed.
Conclusions:
- The novel heterogeneous neural network offers a robust solution for retinal vessel segmentation.
- This approach enhances the diagnostic capabilities for hypertensive retinopathy.
- The model shows promise for clinical applications in automated retinal disease screening.
Abstract:
Retinal vessels play a pivotal role as biomarkers in the detection of retinal diseases, including hypertensive retinopathy. The manual identification of these retinal vessels is both resource-intensive and time-consuming. The fidelity of vessel segmentation in automated methods directly depends on the fundus images' quality. In instances of sub-optimal image quality, applying deep learning-based methodologies emerges as a more effective approach for precise segmentation. We propose a heterogeneous neural network combining the benefit of local semantic information extraction of convolutional neural network and long-range spatial features mining of transformer network structures. Such cross-attention network structure boosts the model's ability to tackle vessel structures in the retinal images. Experiments on four publicly available datasets demonstrate our model's superior performance on vessel segmentation and the big potential of hypertensive retinopathy quantification.
