The adoption of deep learning interpretability techniques on diabetic retinopathy analysis: a review

Abstract

Diabetic retinopathy (DR) is a chronic eye condition that is rapidly growing due to the prevalence of diabetes. There are challenges such as the dearth of ophthalmologists, healthcare resources, and facilities that are unable to provide patients with appropriate eye screening services. As a result, deep learning (DL) has the potential to play a critical role as a powerful automated diagnostic tool in the field of ophthalmology, particularly in the early detection of DR when compared to traditional detection techniques. The DL models are known as black boxes, despite the fact that they are widely adopted. They make no attempt to explain how the model learns representations or why it makes a particular prediction. Due to the black box design architecture, DL methods make it difficult for intended end-users like ophthalmologists to grasp how the models function, preventing model acceptance for clinical usage. Recently, several studies on the interpretability of DL methods used in DR-related tasks such as DR classification and segmentation have been published. The goal of this paper is to provide a detailed overview of interpretability strategies used in DR-related tasks. This paper also includes the authors’ insights and future directions in the field of DR to help the research community overcome research problems. Graphical abstract

Explainable time-series deep learning models for the prediction of mortality, prolonged length of stay and 30-day readmission in intensive care patients

“…To the best of our knowledge, no studies have ever predicted common outcomes while maximizing the value of these models of patients in general ICU. Furthermore, because of the complexity of these deep learning models, they are not easy to interpret, which restricts their practical application to clinical decisions ( 21 , 22 ). Therefore, transparency and explainability must be considered when constructing prediction models.…”

Explainable artificial intelligence in ophthalmology

“…Among XAI methods applied to these AI systems, most employ attribution-based methods to generate posthoc local heatmaps to represent regions of the input image that contribute most to output decision [33,34 ▪ ,35]. By visualizing that attention areas of DL models correlate to clinically relevant features, these visualization heatmaps can potentially boost clinicians’ confidence in model output decisions [17 ▪▪ ,36].…”

Detection of Diabetic Retinopathy using Convolutional Neural Networks for Feature Extraction and Classification (DRFEC)

“…The model introduces the significance of data augmentation in better model performance. Lim et al [ 55 ] have performed a literature review on gradient-based interpretability methods in DL models such as saliency map, integrated gradient, layer-wise relevance propagation, occlusion testing, sensitivity analysis, class activation map, gradient-weighted class activation map and layer-wise relevance propagation, in the detection of DR. It identifies the drawbacks of these interpretability methods in detection of correct lesions for a given class and the lack of reliable ground truth.…”