560-P: Thermal Indicators of Previously Undiagnosed Diabetic Peripheral Neuropathy

Abstract

Purpose: Neurological clinical exams, quantitative sensory tests (QST), and nerve conduction studies (NCS) have been reported to have low accuracy in diagnosing diabetic peripheral neuropathy (DPN). This study explores the use of infrared (thermal) video imaging to detect small and large fiber DPN. Methods: 48 subjects with and 148 subjects without DPN were studied and compared to age-matched control subjects (n=84). All subjects’ feet were examined for sores, deformities, and other signs of disease, and patients with recent foot surgeries were excluded. Monofilament and vibration threshold testing ware performed on all subjects. Thermal videos were collected for all subjects, and NCS exams were performed on a subset of subjects. Thermal imaging followed a standard procedure: Imaging pre- and post-cooling to 2°C. A thermal classification model using thermal features extracted from the videos was developed with NCS as the reference standard to assess the performance of thermoregulation models. Results: Monofilament and vibration test results were compared with clinical diagnoses and had sensitivities of 52% and 57%, respectively. Sensitivity and specificity increase to 74% and 85% when the two tests were combined in an “either/or” decision. NCS was performed on 39 subjects: 11 control subjects, 12 subjects with clinical DPN, and 16 subjects with DM but without DPN. QST demonstrated a best sensitivity of 59% for the vibration test and a best individual specificity of 100% for the monofilament test when compared to NCS. When combined, QST demonstrated best overall sensitivity and specificity of 63% and 87%. The thermal model returned a sensitivity/specificity of 82%/94% and 91%/75% for different cut-points. Conclusions: Thermal recovery modeling from a cold stimulus yielded superior agreement with clinical and NCS diagnoses as compared to QST in the detection of DPN. A thermal imaging device can be made that is feasible for use in clinics at a lower cost than NCS devices. Disclosure M.R. Burge: None. C.S. Calder: None. J. Wigdahl: None. V. Joshi: None. X. Guo: None. J.C. Carmichael: None. P. Soliz: Stock/Shareholder; Self; VisionQuest Biomedical LLC. Funding National Institutes of Health