Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Safety in Diabetic Retinopathy Detection: Uncertainty-Aware Deep Learning Models with Rejection Capabilities

Madhushan Ramalingam, Yaish Riaz, Priyanthi Rajamanoharan, Piyumi Dasanayaka

TL;DR

The paper tackles overconfidence in diabetic retinopathy detection by incorporating a Variational Bayesian Linear Layer to quantify predictive uncertainty. A confidence-based rejection mechanism defers low-confidence cases to clinicians, enabling safer triage. Empirical results show 89.93% accuracy on accepted cases with 74.5% coverage, a 25.5% rejection rate, and an ECE of 0.0217, illustrating a favorable accuracy-calibration trade-off. This approach advances trustworthy AI in healthcare by coupling high-confidence predictions with a safe deferral strategy that can be integrated into real-world clinical workflows.

Abstract

Diabetic retinopathy (DR) is a major cause of visual impairment, and effective treatment options depend heavily on timely and accurate diagnosis. Deep learning models have demonstrated great success identifying DR from retinal images. However, relying only on predictions made by models, without any indication of model confidence, creates uncertainty and poses significant risk in clinical settings. This paper investigates an alternative in uncertainty-aware deep learning models, including a rejection mechanism to reject low-confidence predictions, contextualized by deferred decision-making in clinical practice. The results show there is a trade-off between prediction coverage and coverage reliability. The Variational Bayesian model adopted a more conservative strategy when predicting DR, subsequently rejecting the uncertain predictions. The model is evaluated by means of important performance metrics such as Accuracy on accepted predictions, the proportion of accepted cases (coverage), the rejection-ratio, and Expected Calibration Error (ECE). The findings also demonstrate a clear trade-off between accuracy and caution, establishing that the use of uncertainty estimation and selective rejection improves the model's reliability in safety-critical diagnostic use cases.

Enhancing Safety in Diabetic Retinopathy Detection: Uncertainty-Aware Deep Learning Models with Rejection Capabilities

TL;DR

The paper tackles overconfidence in diabetic retinopathy detection by incorporating a Variational Bayesian Linear Layer to quantify predictive uncertainty. A confidence-based rejection mechanism defers low-confidence cases to clinicians, enabling safer triage. Empirical results show 89.93% accuracy on accepted cases with 74.5% coverage, a 25.5% rejection rate, and an ECE of 0.0217, illustrating a favorable accuracy-calibration trade-off. This approach advances trustworthy AI in healthcare by coupling high-confidence predictions with a safe deferral strategy that can be integrated into real-world clinical workflows.

Abstract

Diabetic retinopathy (DR) is a major cause of visual impairment, and effective treatment options depend heavily on timely and accurate diagnosis. Deep learning models have demonstrated great success identifying DR from retinal images. However, relying only on predictions made by models, without any indication of model confidence, creates uncertainty and poses significant risk in clinical settings. This paper investigates an alternative in uncertainty-aware deep learning models, including a rejection mechanism to reject low-confidence predictions, contextualized by deferred decision-making in clinical practice. The results show there is a trade-off between prediction coverage and coverage reliability. The Variational Bayesian model adopted a more conservative strategy when predicting DR, subsequently rejecting the uncertain predictions. The model is evaluated by means of important performance metrics such as Accuracy on accepted predictions, the proportion of accepted cases (coverage), the rejection-ratio, and Expected Calibration Error (ECE). The findings also demonstrate a clear trade-off between accuracy and caution, establishing that the use of uncertainty estimation and selective rejection improves the model's reliability in safety-critical diagnostic use cases.

Paper Structure

This paper contains 14 sections, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Issues with Current Screening Models
  3. Deep Learning in DR Detection: Progress and Limitations
  4. The Role of Uncertainty Quantification in Medical AI
  5. Literature Review
  6. Methodology
  7. Dataset Preparation
  8. DRD Model with VBLL
  9. Uncertainty-Based Rejection
  10. Evaluation Process
  11. Implementation Details
  12. Results
  13. Discussion
  14. Conclusion

Figures (2)

  • Figure 1: Confidence Distribution
  • Figure 2: Confusion matrix comparison