Anomalous contra-lateral radiometric asymmetry in the diabetic patient

Abstract

We introduce a methodology for acquisition and analysis of infrared (IR) images, picturing the metabolic heat emission from the human skin. Then we analyze the radiometric asymmetries in the patients with DM2 in comparison to the natural asymmetries, represented by the control group. In this regard, we introduce three indices (TAI, SAI, TtAI) with conditions for disclosing asymmetries displayed on images acquired in passive mode (the natural thermal emission, NTE). Then, the indices are adapted for analysis of IR-images acquired in what we brand as active mode (the NTE is altered by means of a controlled external stimulus). Out of the passive mode, the TAI and TtAI indices show the best diagnostic performance, with values of sensitivity and specificity of 89% and 72%, and 83% and 78%, respectively. Instead, from the active mode analysis we get 86% of sensitivity and 83% of specificity for the TRI index. We report data obtained form IR-images of 36 patients with Diabetes Mellitus Type II (DM2) and 18 non-diabetic controls. For both groups the image acquisition is made in passive and active mode, picturing the anterior and posterior views of the lower limbs. With this analysis, we manage to unveil the contra-lateral radiometric asymmetries of the legs, along with the differences between patients and controls. Finally, we report the consistency of these indices with glucose and glycated hemoglobin (HbA1c), known to be the golden clinical variables used to diagnose DM2.

Exploring the effect of environmental thermal noise in the contra-lateral thermal asymmetry of the lower limbs

“…According with the ROC-analysis, except for the TAI, that conserves its diagnostic performance (89% of sensitivity and 72% of specificity) with and without noise, indices SAI and TtAI change their performances as follows: for the SAI the specificity increases from 67% to 72%, and the sensitivity remains in 78%; in the case of the TtAI the specificity remains in 78% and the sensitivity increases from 83% to 88% [5].…”

“…The task is to achieve a large and reliable IRI data base. The details on the methodology for the thermal asymmetry study are reported in [5]. In the current work we produce a comparative study, by getting information from a region of interest (ROI) of the IRIs.…”

“…The next step is to get the relative thermal asymmetry between the lower limbs. For this purpose, we use the procedure and indices introduced in [5]. In it each image pixel represents an actual temperature value.…”

“…Under such conditions the lower extremities would display altered contra-lateral thermal asymmetry, respect of what could be the natural asymmetry. Details on the subject have been reported in [5].…”

“…For the matter, we analyze the effect of local environment fluctuations, which are commonly neglected at time of analyzing IRIs. The general assessment is to assume the environment as stable under controlled room temperature, using domestic-like coolant systems [5]. For biomedical uses of IRI, such analysis is not good enough, and more accurate information is not available in a practical sense.…”