Diabetes Mellitus (DM) is a huge public health-care chronic disease issue, with about 415 millions of patients worldwide in 2015. DM is a serious pathology, especially through its complications and their severity, as one, called diabetic foot, which can lead to a lower limb amputation. The aim of the present study is to propose a dynamic method to classify the foot state through a thermal method. This one uses infrared imaging recorded after the application of a local cold stimulus and a bio-heat model. The model's thermophysical and micro-perfusion parameters are identified and compared to the patients' clinical classification.