Modeling Thermal Infrared Imaging Data for Differential Diagnosis

Abstract

Nowadays, thermal infrared imaging (IRI) is thought to be a fascinating and promising complementary imaging tool regarding typical gold-standard medical imaging for differential diagnosis. This chapter presents the commonly used approaches for modeling thermal infrared data for differential diagnosis purposes. Two main modeling approaches were proposed, i.e., (i) qualitative modeling approach based on using statistical and machine learning techniques, (ii) quantitative modeling approach based on performing mathematical/analytical modeling of the thermoregulatory processes by using three main approaches: (i) empirically using automatic control theory, (ii) non-empirically using bioheat equations and (iii) semi-empirically using both bioheat equations and automatic control theory. Also, three main modeling approaches based on control system theory were presented, i.e., (i-a) time domain analysis of the thermoregulatory system’s characteristics through a direct estimation of the closed loop dynamic response parameters of a prototype second-order system, (i-b) a direct identification of thermoregulatory system as a second-order system plus delay time (SOPDT) from a closed-loop step response, and (i-c) a state-space representation of the thermoregulatory system as a first-order differential equation from the experimental IR temperature curves. Moreover, this chapter summarizes the advantages and disadvantages of each modeling approach highlighting its assumptions and approximations. By implementing the proposed modeling approaches, thermal infrared imaging has been demonstrated to be able to (i) identify significant averaged and asymmetric temperature parameters that could be used for disease classification, (ii) provide a direct functional IR indicators of the thermoregulatory malfunctions/alternations indirectly assess the severity of functional perturbation of the autonomic sympathetic and parasympathetic physiological activations in the presence of a disease, (iii) compute physiological information, such as localized blood flow, cardiac pulse, and breath rate, and (iv) identify skin’s thermal parameters, location of heat source (particularly the vessels), depth of heat source used for defining the location and geometrical shape of the affected-area, mostly required for tumor detection, and (v) provide a clear description of the underlying alterations in the main thermoregulatory functions as for example, environmental heat exchange process, vasoconstriction and/or vasodilation, and sweating actions. The authors consider this chapter as a good material that provides a great insight about the utility of thermal infrared imaging for medical diagnostic purposes.