Correlation of Dermal Features with Diabetes Stage

This article presents a research study that investigates the correlation between dermal features derived from optoacoustic tomograms and diabetes stages using machine learning techniques. The study involved obtaining 199 raster-scan optoacoustic mesoscopy (RSOM) images from 115 participants, including 40 healthy individuals and 75 diagnosed with diabetes. The researchers utilized machine learning to segment skin layers and microvasculature, identifying features that predict diabetes presence with significant accuracy. A notable finding was the development of a 'microangiopathy score' based on 32 relevant features, which demonstrated an area under the receiver operating characteristic curve of 0.84. The study emphasizes the potential of RSOM in discovering biomarkers for diabetes and monitoring its status through non-invasive methods. It also discusses the importance of understanding microvascular changes in the skin as indicators of diabetes progression, which could enhance clinical practices in assessing the disease.