Medical thermography: a diagnostic approach for type 2 diabetes based on non-contact infrared thermal imaging

Abstract

To test the potential of Infrared (IR) thermography in diagnosing as well as predicting type 2 diabetes and its complications compared with biochemical assay of HbA(1c) as standard. As per American Diabetes Association criteria, threshold for diagnosis of diabetes was set as HbA(1c) ≥ 6.5 % (7.7 mmol L⁻¹). The total subjects (n = 62) were studied out of which control (n = 32) and diabetic subjects (n = 30). IR camera was used to capture the thermal images of the skin for diagnosis of the disease; receiver operating characteristic (ROC) curve was used to set temperature (°C) as threshold for statistically significant body regions under t test. In diabetic group, HbA(1c) showed negative correlation with carotid region (r = -0.471, p < 0.01) and the mean skin temperature was lower than the normal group at body regions namely knee (p = 0.002), tibia (p = 0.003), forehead (p = 0.014), and palm (p = 0.019). The palm region showed highest area under the curve of 0.711 (95% CI: 0.581-0.842) and the threshold was set as ≤33.85 °C, thereby sensitivity (90%) and specificity (56%) was obtained in determining the undiagnosed diabetes with positive predictive value of 65%, negative predictive value of 85% and accuracy of 73%. As HbA(1c) increases, skin temperature decreases. Skin temperature enables early detection of diabetes as compared to HbA(1c). The decrease in skin temperature may be due to the decrease in the basal metabolic rate, poor blood perfusion and high insulin resistance. Thermography can be used as a diagnostic as well as prognostic tool for the diabetes.