An intelligent telemedicine system for detection of diabetic foot complications

Early identification and timely treatment of diabetic foot complications are essential in preventing their devastating consequences such as lower-extremity amputation and mortality. Frequent and automatic risk assessment by an intelligent telemedicine system may be feasible and cost-effective. As the first step to approach such a telemedicine system, an experimental setup that combined three promising imaging modalities, namely spectral imaging, infrared thermal imaging, and photometric stereo imaging, was developed and investigated. The spectral imaging system in the experimental setup contains nine cameras in a matrix configuration, fitted with the preselected optical filters. Using the spectral images acquired, front-end pixel classifiers were developed to detect the diabetic foot complications automatically. Taking the image annotations based on live assessment as ground truth, the validation results indicate that these front-end classifiers can identify the diabetic foot complications with acceptable performance. However, future studies are needed on enhancing the performance of current pixel classifiers and designing the back-end classifiers. With the infrared thermal imaging, images of temperature distributions can be acquired from patients’ feet. The temperature differences between the corresponding areas of the contralateral feet are clinically significant parameters for identifying the diabetic foot complications. To detect this temperature differences automatically, an asymmetric analysis were proposed and investigated. Results show that the corresponding points on the two feet can be identified irrespective of the shapes, sizes or poses of the feet. With the photometric stereo imaging, a feasibility study were conducted to detect diabetic foot complications with the 3D surface reconstruction. The results indicate that this imaging technology may be promising but subjected to some limitations currently, such as the movement in patients' foot during image acquisition. To determine the potential value of this modality in the future telemedicine system, further improvement is required. The outcomes of the studies presented in this thesis showed the feasibility of developing a telemedicine system to detect diabetic foot complications with the three imaging modalities. The studies acted as the precursors for developing an intelligent telemedicine system, which proposed potential detection methodologies and provided the directions for the future study.