Identifying diabetes from conjunctival images using a novel hierarchical multi-task network

Abstract

Diabetes can cause microvessel impairment. However, these conjunctival pathological changes are not easily recognized, limiting their potential as independent diagnostic indicators. Therefore, we designed a deep learning model to explore the relationship between conjunctival features and diabetes, and to advance automated identification of diabetes through conjunctival images. Images were collected from patients with type 2 diabetes and healthy volunteers. A hierarchical multi-tasking network model (HMT-Net) was developed using conjunctival images, and the model was systematically evaluated and compared with other algorithms. The sensitivity, specificity, and accuracy of the HMT-Net model to identify diabetes were 78.70%, 69.08%, and 75.15%, respectively. The performance of the HMT-Net model was significantly better than that of ophthalmologists. The model allowed sensitive and rapid discrimination by assessment of conjunctival images and can be potentially useful for identifying diabetes.

Intellectual System Based on Trained Neural Network Models in Assesment of Blood Microcirculation Disturbances Using Bulbar Conjunctival Images.

“…На сегодняшний день имеется ряд работ по применению ИИ для оценки микроциркуляции в конъюнктиве глаза, из которых большинство посвящено изучению гемодинамики [23,24]. Также предложена модель ИИ для распознавания сахарного диабета по изображениям БК [25].…”

Progress of Bulbar Conjunctival Microcirculation Alterations in the Diagnosis of Ocular Diseases

“…The gold standard for DR diagnosis is ophthalmic fundus examination, which often requires dilated pupils and is not suitable for large-scale screening [ 31 ]. Alterations in the bulbar conjunctival microcirculation reflect the severity of diabetic vasculopathy, with sensitivity and specificity as diagnostic indicators of type 2 diabetes reaching 78.70% and 69.08% [ 32 ]. And its observation has the advantages of simple operation, short time consumption, and no need to pretreat patients, and it is promising to be applied to the early diagnosis or large-scale screening of diabetic retinopathy.…”