Development of an earlobe crease identification system to support diagnosis of arteriosclerosis.

It is well known that an earlobe crease (ELC) is related with cardiovascular disease. However evaluation of ELC has been carried out subjectively. We propose a novel system for objective and quantitative evaluation of ELC characteristics. The outline of earlobe was detected using the Canny edge detector. The number of ELCs and the ratio of the ELC length for the lobule from photographs of bilateral earlobes were calculated from the extracted outline of earlobes. Our system's validity for ELCs detection was compared with doctors' diagnosis using sensitivity and specificity analysis. The result showed that the validity of our system was equal to the diagnosis of the medical doctors (kappa coefficient: 0.93). It was thus concluded that the proposed system is useful to predict atherosclerotic disease.