Skin Cancer Classification Using Extended 5 Channel (I-RGB-U) Images Generated From RGB Images

This paper presents a method to make a diagnosis of skin cancer using hyperspectral (HS) images generated from RGB images. To address this issue, we utilize ‘I-RGB-U’ images, which is composed of I (infrared) channel and U (ultraviolet) channel as well as RGB channels. In recent years, the number of skin cancer patients is increasing, and early detection becomes more and more important. HS cameras have attracted attention for diagnosing skin diseases due to their ability to capture fine spectral information of the skin, but they are too expensive to widely adopt in households. The purpose of this paper is to generate I-RGB-U images from RGB images without using HS cameras, and to enhance the performance of skin cancer diagnosis by using them as input.