Deep learning differentiates between healthy and diabetic mouse ears from optical coherence tomography angiography images

Abstract

We trained a deep learning algorithm to use skin optical coherence tomography (OCT) angiograms to differentiate between healthy and type 2 diabetic mice. OCT angiograms were acquired with a custom-built OCT system based on an akinetic swept laser at 1322 nm with a lateral resolution of ∼13 μm and using split-spectrum amplitude decorrelation. Our data set consisted of 24 stitched angiograms of the full ear, with a size of approximately 8.2 × 8.2 mm, evenly distributed between healthy and diabetic mice. The deep learning classification algorithm uses the ResNet v2 convolutional neural network architecture and was trained on small patches extracted from the full ear angiograms. For individual patches, we obtained a cross-validated accuracy of 0.925 and an area under the receiver operating characteristic curve (ROC AUC) of 0.974. Averaging over multiple patches extracted from each ear resulted in the correct classification of all 24 ears.

Classification of Tear Film Lipid Layer En Face Maps Obtained Using Optical Coherence Tomography and Their Correlation With Clinical Parameters

“…Both aspects could be improved by automating the image analysis, either by traditional feature extraction or with convolutional neural networks, of which the latter are perfectly suited for image classification tasks. 48 Nevertheless, the high ICC shows that the selected pattern classes are easily distinguishable. An additional disadvantage of the current data is that the LLT values and TFLL pattern images are not acquired exactly at the same time.…”

A Decision Support System for Diagnosing Diabetes Using Deep Neural Network

“…The training dataset was used to build a model for predicting risk based on information gathered at registration. The deep learning classification technique makes use of the ResNet v2 CNN architecture ( 24 ), which was trained on tiny patches taken from the entire ear endoscopies before being applied to the complete ear images. A total of four deep learning models were trained for autonomous ascribable diabetic retinopathy detection, dependent on whether or not two criteria were included: DR-related lesions and diabetic retinopathy staging ( 25 ).…”