Classification of Skin Cancer: ANN Trained with Scaled Conjugate Gradient Algorithm

Abstract

Automatic detection and accurate classification of skin cancer has been achievable with the help of machine learning techniques. Now-a-days, skin cancer has posed to be a dreadful disease mostly seen in the western region of the world. The current work focuses on the automated identification of skin cancer. The proposed method initially deals with pre-processing steps like normalization and image enhancement. Texture based different features of the digital skin images are extracted by using Gray Level Co-occurrence Matrices (GLCM). Then classification is performed based on the extracted features using Support Vector Machine (SVM), Ensemble and Artificial Neural Network (ANN). To train the ANN, Scaled conjugate gradient (SCG) algorithm has been adopted. This paper compares the performance of these three classifiers in terms of some parameters namely accuracy, precision, recall and f-measure. The experimental results show the effectiveness of ANN-SCG classifier. The ANN-SCG model provides 84.21% accuracy, 83.33% precision, 71.42% f-measure and 76.91% recall which is superior to the SVM and Ensemble classifiers.