Extreme Learning Machine for Melanoma Classification

Melanoma is considered the most dangerous type of skin cancer worldwide. Early detection of this cancer type plays an important role in the treatment process. Many invasive methods such as shaving were used for taking a biopsy from the patients for diagnosis. However, nowadays image processing techniques combined with machine learning can play a significant role in the diagnosis of medical images in a non-invasive way. To this end, this study aims to use machine learning algorithms for the early diagnosis and treatment of this type of cancer. Methodologically, the steps involved in this study are preprocessing of melanoma images, segmentation, features extraction, and then classification using the extreme learning machine (ELM) classifier. To the best of our knowledge, this classifier has not been used before for such types of images. This study aims to test the efficacy of using ELM in classifying these types of images compared with other common types of Machine learning algorithms such as Random Forest, support vector machine (SVM), and K-nearest neighbor (KNN). Promising results were obtained from ELM with an accuracy of 97%, 91 % using 11 features, and 5 features, respectively.