Preliminary Melanoma Detection Mobile Application using Support Vector Machine Classification

This paper proposes a mobile application that uses a mobile phone camera attached to an enhanced lens to capture images of any suspicious portrusions on the body (e.g. mole) and be able to predict whether it is melanoma using image processing and machine learning techniques. The images are preprocessed to remove the noise and segment the region of interest (ROI). Features that distinguish melanoma from normal tissues are extracted such as the texture, color, and geometrical shape. The proposed method uses Support Vector Machine (SVM) classification algorithm for training and prediction. The proposed method is implemented and tested on publicly available datasets. Experimantal results showed that the method was able to detect the melanoma cases with a prediction accuracy of 79%.