Artificial intelligence utilising corneal confocal microscopy for the diagnosis of peripheral neuropathy in diabetes mellitus and prediabetes – published online 21/11/2021

Frank G. Preston, Yanda Meng, Jamie Burgess, Maryam Ferdousi, Shazli Azmi, Ioannis N. Petropoulos, Stephen Kaye, Rayaz A. Malik, Yalin Zheng, Uazman Alam

The accurate detection of diabetic neuropathy in routine clinical practice remains a major unmet clinical need. Current screening practices largely rely on insensitive tests which only primarily detect the insensate foot. It has been demonstrated that artificial intelligence (AI) trained using annotated corneal confocal microscopy images can provide accurate segmentation of corneal nerve images, allowing the detection of peripheral neuropathy. In this issue, Preston and Meng et al (https://doi.org/10.1007/s00125-021-05617-x) report that AI utilising corneal nerve images can accurately classify peripheral neuropathy in people with prediabetes and diabetes, without the need for underlying nerve segmentation. This was achieved by use of a single corneal confocal microscope image. The authors discuss that, as annotation of the image/dataset was not required, larger sets of unannotated images may be leveraged in providing a more robust model. The authors conclude that with validation in a larger real-world study, the AI algorithm has considerable potential for adoption into screening programmes for diabetic neuropathy.

All News