Detection of diabetic peripheral neuropathy using spatial-temporal analysis in infrared videos

摘要

Limitations of previous thermographic studies to detect diabetic peripheral neuropathy (DPN) are addressed in this combined analysis using spatial and temporal features. In our approach, we extract information of temperature patterns before cooling and during recovery after cooling. Temporal features are extracted from, angiosome patterns, principal component analysis (PCA) and independent component analysis (ICA) from the recovery stage after applying a cold stimulus to the plantar foot. The features are processed by a linear support vector machine (SVM) classifier achieving area under the ROC curve (AUC) of 0.95 and 0.83 for the detection of DPN in females and males respectively.