Abstract
Purpose :
To investigate the clinical association of diabetic retinopathy (DR) and retinal vascular abnormalities, with microvascular function in the plantar foot as recorded in thermal recovery characteristics in diabetic peripheral neuropathy (DPN).
Methods :
A clinical study was performed for a cohort of 20 controls and 10 subjects with diagnosed DPN (50-75 years of age). All subjects underwent a clinical examination for DPN, retinal imaging, and thermal imaging of the plantar foot. The retinal images were captured with Canon CR2 camera and graded by a certified retinal reader for five DR severity levels; and for vessel abnormalities: tortuosity, artery-vein ratio, artery-venous nicking, branching angle, vessel caliber. Thermal imaging protocol for the plantar feet was: 1) pre-cooling stage, 2) application and subsequent removal of cold provocation, 3) recovery of plantar foot temperature toward normal state. Thermal features were: post cooling temperature drop, temperature recovery rate, spatial patterns in thermal recovery, and thermal regulation characteristics. The statistical analysis used T-test, Chi-Square test, and logistic regression techniques to determine significance of retinal and thermal features in classifying DPN and control groups.
Results :
A T-test was performed to discover retinal features distinguishing DPN from controls. The most significant features were the DR levels and number of arterial-venous nicking (p < 0.05). Furthermore, a logistic regression classifier was developed using DR levels and arterial-venous nicking; that demonstrated statistical significance (p < 0.05), as well as an accuracy of 0.82 in classifying subjects with DPN and controls. Similarly, the association between retinal and thermal features was demonstrated by statistically significant difference in temperature recovery rates measured at the ball area of both feet, for subjects with DR/artery-venous nicking and subjects without these retinal abnormalities (p < 0.05). These thermal features also showed an accuracy of 0.72 in classifying subjects with DPN and controls.
Conclusions :
As the diabetic patients are recommended to undergo DR screening every year, the proposed system can be seamlessly integrated with DR screening at no extra cost. This research demonstrates an added benefit of retinal screening by identifying patients with undiagnosed DPN, enabling timely treatment.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.