Artificial intelligence utilising corneal confocal microscopy for the diagnosis of peripheral neuropathy in diabetes mellitus and prediabetes

Abstract

Aims/hypothesis We aimed to develop an artificial intelligence (AI)-based deep learning algorithm (DLA) applying attribution methods without image segmentation to corneal confocal microscopy images and to accurately classify peripheral neuropathy (or lack of). Methods The AI-based DLA utilised convolutional neural networks with data augmentation to increase the algorithm’s generalisability. The algorithm was trained using a high-end graphics processor for 300 epochs on 329 corneal nerve images and tested on 40 images (1 image/participant). Participants consisted of healthy volunteer (HV) participants (n = 90) and participants with type 1 diabetes (n = 88), type 2 diabetes (n = 141) and prediabetes (n = 50) (defined as impaired fasting glucose, impaired glucose tolerance or a combination of both), and were classified into HV, those without neuropathy (PN−) (n = 149) and those with neuropathy (PN+) (n = 130). For the AI-based DLA, a modified residual neural network called ResNet-50 was developed and used to extract features from images and perform classification. The algorithm was tested on 40 participants (15 HV, 13 PN−, 12 PN+). Attribution methods gradient-weighted class activation mapping (Grad-CAM), Guided Grad-CAM and occlusion sensitivity displayed the areas within the image that had the greatest impact on the decision of the algorithm. Results The results were as follows: HV: recall of 1.0 (95% CI 1.0, 1.0), precision of 0.83 (95% CI 0.65, 1.0), F1-score of 0.91 (95% CI 0.79, 1.0); PN−: recall of 0.85 (95% CI 0.62, 1.0), precision of 0.92 (95% CI 0.73, 1.0), F1-score of 0.88 (95% CI 0.71, 1.0); PN+: recall of 0.83 (95% CI 0.58, 1.0), precision of 1.0 (95% CI 1.0, 1.0), F1-score of 0.91 (95% CI 0.74, 1.0). The features displayed by the attribution methods demonstrated more corneal nerves in HV, a reduction in corneal nerves for PN− and an absence of corneal nerves for PN+ images. Conclusions/interpretation We demonstrate promising results in the rapid classification of peripheral neuropathy using a single corneal image. A large-scale multicentre validation study is required to assess the utility of AI-based DLA in screening and diagnostic programmes for diabetic neuropathy. Graphical abstract

Multi-Model Domain Adaptation for Diabetic Retinopathy Classification

“…Retinal images are usually used to build an automatic diabetic retinopathy diagnosis system (Gardner et al, 1996;Acharya et al, 2009; Ram et al, 2010;Gulshan et al, 2016;Lam et al, 2018;Jiang et al, 2019;Preston et al, 2021). However, whether using traditional machine learning methods (Gardner et al, 1996;Acharya et al, 2009;Ram et al, 2010) or deep supervised learning methods (Gulshan et al, 2016;Lam et al, 2018;Jiang et al, 2019;Preston et al, 2021), they all need a large amount of labeled data during training. In the biomedical image analysis field, labeling work is expensive and the privacy issue is highly sensitive.…”

“…Artificial intelligence (AI) is a popular technique for computer-aided automatic DR diagnosis to overcome these obstacles and deep learning has achieved progress in biomedical image analysis (Meng et al, 2021b;Preston et al, 2021;Meng et al, 2021a). Yoo and Park (2013) utilized ridge, elastic net, and LASSO to perform validation on 1052 DR patients.…”

Artificial Intelligence and Corneal Confocal Microscopy: The Start of a Beautiful Relationship

“…Subsequently, Chen et al [ 81 ] demonstrated comparable efficacy between ACCMetrics and IENFD in diagnosing DPN. Recent advances in the automated analysis of CCM images have utilised DLA [ 88 , 89 , 90 ] and abandoned nerve segmentation in favour of end-to-end classification to allow the DLA to determine features of importance for image classification [ 91 , 92 ].…”

“…Mou et al [ 125 ] developed a corneal nerve fibre segmentation model based on a dual attention (e.g., spatial and channel attentions) mechanism for the prediction of regions of interest and demonstrated the model’s effectiveness through their automated DLA that demonstrated significant differences in the tortuosity of nerves between patients with diabetes and healthy controls [ 124 ]. Preston et al [ 92 ] developed a DLA using ResNet [ 126 ] as the backbone network to diagnose peripheral neuropathy (in diabetes and prediabetes), achieving a high level of classification accuracy using end-to-end classification ( Figure 2 ). They also demonstrated image attribution-based explainability methods to produce ‘heatmaps’ showing areas in the image which were important to the classification prediction ( Figure 3 ).…”

The Validation of Deep Learning-Based Grading Model for Diabetic Retinopathy

“…DL-based technology has been applied to retinal vascular segmentation, recognition, and classification of DR lesions, referable DR and diabetic neuropathy detection. Recently, Preston et al developed an AI-based algorithm to classify peripheral neuropathy utilizing CCM without image segmentation prior to classification, which did not require manual or automated annotation and allowed the utilization of a larger database ( 13 ). Most of the above algorithms use the convolutional neural networks (CNN) architecture, which has better performance than other network architectures ( 14 ).…”