Related Experiment Videos
Integrating prior knowledge with deep learning for optimized quality control in corneal images: A multicenter study
Fen-Fen Li1, Gao-Xiang Li2, Xin-Xin Yu1
1National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, PR China.
Computer Methods and Programs in Biomedicine
|May 4, 2025
Summary
A new Hybrid Prior-Net (HP-Net) system effectively classifies slit-lamp images, improving diagnostic accuracy in telemedicine. This AI tool enhances medical imaging quality control by handling image variability.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Artificial intelligence (AI) models excel with high-quality slit-lamp images but struggle with real-world clinical variability.
- Image quality control is crucial for reliable AI-driven diagnostics in ophthalmology.
- Telemedicine applications require robust AI tools that can handle diverse image conditions.
Purpose of the Study:
- To develop and evaluate a hybrid AI-based image quality control system for classifying slit-lamp images.
- To enhance diagnostic accuracy and efficiency in ophthalmic telemedicine.
- To address challenges posed by image variability in clinical settings.
Main Methods:
- A cross-sectional study utilizing internal (Zhejiang Eye Hospital, 2982 images) and external datasets (Aier Guangming Eye Hospital, 13,554 images; First People's Hospital of Aksu, 9853 images).
- Development of Hybrid Prior-Net (HP-Net), a novel network combining ResNet classification with Hough circle transform and frequency domain blur detection.
- Feature concatenation in HP-Net to improve classification of eligible, misaligned, blurred, and underexposed corneal images.
Main Results:
- HP-Net achieved superior performance with 99.03% accuracy, 98.21% precision, 95.18% recall, 99.36% specificity, and 96.54% F1-score.
- HP-Net effectively filtered images from external datasets, achieving 97.23% (AGEH) and 96.97% (FPH of Aksu) accuracy.
- Demonstrated superior feature extraction and classification capabilities of HP-Net across all metrics.
Conclusions:
- The AI-based image quality control system provides a robust solution for corneal image classification, significantly benefiting telemedicine.
- Incorporating usable but slightly blurred images into training enhances AI reliability and adaptability for medical imaging quality control.
- The system paves the way for more accurate and efficient diagnostic workflows in ophthalmology.
Related Concept Videos
02:07Improving Translational Accuracy
2.5K
2.5K
01:15Depth Perception and Spatial Vision
499
Depth perception is the ability to perceive objects three-dimensionally. It relies on two types of cues: binocular and monocular. Binocular cues depend on the combination of images from both eyes and how the eyes work together. Since the eyes are in slightly different positions, each eye captures a slightly different image. This disparity between images, known as binocular disparity, helps the brain interpret depth. When the brain compares these images, it determines the distance to an object.
499
01:20Deconvolution
123
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
123