Automatic analysis of plantar foot thermal images in at-risk type ii diabetes by using an infrared camera

Abstract

Temperature of the plantar foot surface is an important feature in type II diabetes as abnormal temperature variations can be an early sign of foot diseases. In this paper, automatic way to analyze these temperature variations is presented by using an infrared camera. A robust acquisition protocol is proposed and an image processing software is developed. Three types of analysis are performed. First, the mean plantar foot temperature of both feet results from a segmentation procedure based on the Chan and Vese active contour method. Second, the point-to-point absolute mean difference between the 2 feet is assessed by using a rigid registration method. Third, significant hyperthermia regions such that the point-to-point absolute difference is greater than 2.2°C are highlighted. All these measures are fully automatic and do not need manual intervention. 82 type II diabetic subjects in a preulcerative state were recruited in the Dos de Mayo hospital (HNDM) in Lima, Peru. These persons were classified in two risk groups of developing an ulcer based on a medical exam: a medium risk group, and a high risk group. Results show that the mean temperature of the plantar foot surface is higher of 1ºC in the high risk group compared to the medium risk group. The mean point-to-point absolute difference shows identical values in the 2 groups. Finally, 9 subjects out of the 82 ones show significant hyperthermia of one foot compared to the other (6 in the medium risk group and 3 in the high risk group). It is expected that the new opportunity to automatically analyze foot temperature in hospitals or in diabetic health centers will help in reducing foot ulcer occurrence for type II diabetic persons.