Computer Science ETDs

Viktor Chekh

Publication Date

7-1-2016

Abstract

Diabetes is an incurable metabolic disease characterized by high blood sugar levels. The feet of people with diabetes are at the risk of a variety of pathological consequences including peripheral vascular disease, deformity, ulceration, and ultimately amputation. The key to managing the diabetic foot is prevention and early detection. Unfortunately, current hospital centered reactive diabetes care and the availability of inadequate qualitative diagnostic screening procedures causes physicians to miss the diagnosis in 61% of the patients. We have developed a computer aided diagnostic system for early detection of diabetic foot. The key idea is that diabetic foot exhibits significant neuropathic and vascular damages. When a diabetic foot is placed under cold stress, the thermal recovery will be much slower. This thermal recovery speed can be a quantitative measure for the diagnosis of diabetic foot condition. In our research, thermal recovery of the feet following cold stress is captured using an infrared camera. The captured infrared video is then filtered, segmented, and registered. The temperature recovery at each point on the foot is extracted and analyzed using a thermal regulation model, and the problematic regions are identified. In this thesis, we present our research on the following aspects of the developed computer aided diagnostic systems: subject measurement protocols, a trustful numerical model of the camera noise and noise parameter estimations, infrared video segmentation, new models of thermal regulations, thermal patterns classifications, and our preliminary findings based on small scale clinical study of about 40 subjects, which demonstrated the potential the new diagnostic system.

Language

English

Keywords

Diabetes, Control theory, Thermal infrared, Image processing, Video processing, cold stress, thermal recovery, thermal regulation, thermoregulation, classification, filcker noise, vasoconstriction, vasodilation

Document Type

Dissertation

Degree Name

Computer Science

Level of Degree

Doctoral

Department Name

Department of Computer Science

First Committee Member (Chair)

Estrada-Piedra, Trilce

Second Committee Member

Soliz, Peter

Third Committee Member

Yang, Yin

Project Sponsors

National Institute of Diabetes and Digestive and Kidney Diseases Grant DK093192 National Science Foundation Grant CBET-0853157

Download

Share

COinS