Automated analysis of thermal images for peripheral vascular disease monitoring

Abstract

Diabetes is a significant health problem worldwide with its prevalence having been on the increase for at least the last 30 years and all estimates show that this trend will continue. Diabetic patients are at a higher risk for developing peripheral arterial disease (PAD). PAD is a disease in which plaque build up in the arteries restricts blood ow to the peripheries leading to complications such as ulcerations and amputations in the limbs. This work presents a system for the monitoring of peripheral arterial disease in the lower limbs of diabetic patients using thermal imaging. Thermal data was collected from three different population samples, which include both healthy and diabetic participants. The thermal data consists of images of the volar aspect of the hands, anterior aspect of the shins and dorsal aspect of the foot acquired using a pre-defined acquisition protocol. A set of algorithms were developed with the aim of automatically extracting temperature data from 44 anatomical regions of interest across the three body regions. Analysis of this data may identify relevant patterns of interest which may be used to identify between different sub-groups in the thermal image database collected for this work. Results have shown that the regions of interest are extracted with a high accuracy from the participants in our database. The system also provides standardised and repeatable results, and does so in less time then a manual extraction process would take. This shows that a clinical tool which monitors PAD in diabetic patients based on thermal imaging is possible.