Ophthalmic Disease Detection via Deep Learning With a Novel Mixture Loss Function.

Area of Science:

• Ophthalmology
• Computer-aided diagnosis
• Deep learning

Background:

• Computer-aided diagnosis (CAD) systems are increasingly used for detecting ophthalmic diseases.
• Deep learning methods show promise but struggle with data imbalance and outliers in fundus images.
• Accurate detection of cataract, glaucoma, and age-related macular degeneration (AMD) is crucial.

Purpose of the Study:

• To develop a robust deep learning model for automatic eye disease detection.
• To improve the performance of deep learning models on ophthalmic datasets with class imbalance and outliers.
• To introduce a novel mixture loss function for enhanced classification accuracy.

Main Methods:

• A deep neural network model was developed for analyzing retinal fundus color images.
• A novel mixture loss function combining focal loss and correntropy-induced loss was proposed.
• The model was evaluated on a real-life ophthalmic dataset using metrics like accuracy, sensitivity, specificity, Kappa, and AUC.

Main Results:

• The proposed deep learning model with the mixture loss function demonstrated improved performance compared to baseline models.
• The model showed effectiveness and robustness in detecting eye diseases from fundus images.
• Experimental results validated the superiority of the proposed approach in handling complex datasets.

Conclusions:

• The novel mixture loss function enhances the performance of deep learning models for ophthalmic disease detection.
• The proposed model offers a robust and effective solution for automatic diagnosis of common eye conditions.
• This approach holds potential for improving CAD systems in ophthalmology.