An Internet-based Melanoma Diagnostic System - Toward the Practical Application -

Abstract

In this paper, we report a practical application of world’s first internet-based melanoma diagnostic system. The system is now available from all over the world, 24 hours 365 days. As key components of this system, we developed a new dermatologist-like tumor area extraction algorithm and an artificial neural network (ANN) classifier. Our dermatologist-like tumor area extraction algorithm achieved superior extraction performance and the ANN classifier achieved classification accuracy of 97.3% in sensitivity and 86.1% in specificity with leave-one-out cross-validation test of 319 dermoscopy images. Our system supported SSL encrypted transaction and required only several seconds to complete a procedure. On the other hand, we developed portable skin camera as the alternative of dermoscopy and started field-application tests by distributing them for hospitals or medical universities at first setout for making the system into practical use.