A Novel Adaptive Hybrid Focal-Entropy Loss for Enhancing Diabetic Retinopathy Detection Using Convolutional Neural Networks

2024-11-16

arXiv.org

AI Summary

Severe class imbalance—particularly scarcity of mid-to-late-stage diabetic retinopathy (DR) samples—compromises model sensitivity in multi-level DR classification. Method: We propose an Adaptive Mixed Focal-Entropy Loss (AMFE-Loss), the first to synergistically integrate focal loss and entropy loss. It employs a dynamic weighting mechanism to adaptively amplify learning emphasis on minority and hard-to-classify samples, overcoming performance bottlenecks of standard cross-entropy on imbalanced medical data. AMFE-Loss is integrated with CNN backbones (e.g., ResNet50, DenseNet121), incorporating class recalibration and hard-example focusing. Contribution/Results: On DR grading, ResNet50+AMFE-Loss achieves 99.79% overall accuracy and significantly improves sensitivity for Stage 4—the clinically critical late-stage class—demonstrating superior effectiveness and generalizability for identifying high-risk pathological categories in real-world clinical settings.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Multi-modal Vision
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Diabetic retinopathy is a leading cause of blindness around the world and demands precise AI-based diagnostic tools. Traditional loss functions in multi-class classification, such as Categorical Cross-Entropy (CCE), are very common but break down with class imbalance, especially in cases with inherently challenging or overlapping classes, which leads to biased and less sensitive models. Since a heavy imbalance exists in the number of examples for higher severity stage 4 diabetic retinopathy, etc., classes compared to those very early stages like class 0, achieving class balance is key. For this purpose, we propose the Adaptive Hybrid Focal-Entropy Loss which combines the ideas of focal loss and entropy loss with adaptive weighting in order to focus on minority classes and highlight the challenging samples. The state-of-the art models applied for diabetic retinopathy detection with AHFE revealed good performance improvements, indicating the top performances of ResNet50 at 99.79%, DenseNet121 at 98.86%, Xception at 98.92%, MobileNetV2 at 97.84%, and InceptionV3 at 93.62% accuracy. This sheds light into how AHFE promotes enhancement in AI-driven diagnostics for complex and imbalanced medical datasets.

Problem

• Addressing class imbalance in diabetic retinopathy detection
• Improving model sensitivity for minority severity classes
• Enhancing AI diagnostics with adaptive hybrid loss function

Innovation

• Adaptive Hybrid Focal-Entropy Loss for imbalance
• Combines focal loss and entropy loss adaptively
• Enhances CNN performance in medical diagnostics