IJASD
Cnn-Based Method For Benign And Malignant Skin Lesion Classification
Skin cancer is one of the most common forms of cancer, with its diagnosis typically involving manual screening and visual examination by dermatologists. However, these traditional methods are often complex, time-consuming, and prone to errors. With the growing incidence and significant mortality rates associated with skin cancer, early detection is crucial for effective treatment. In this study, we present an optimized deep convolutional neural network (DCNN) model specifically designed to classify skin lesions as either benign or malignant, thereby improving diagnostic accuracy. Our approach utilizes a dataset of 3,297 dermoscopic images and incorporates rigorous data preprocessing along with the softmax activation function. We evaluated multiple optimizers—Adam, RMSProp, and SGD—each configured with a learning rate of 0.0001. Our findings indicate that the Adam optimizer outperforms others, achieving an accuracy of 84%, a loss rate of 32%, a recall of 85%, a precision of 85%, an F1-score of 85%, and a ROC-AUC of 83% on the ISIC dataset. This CNN-based approach offers a promising tool for enhancing early skin cancer detection and diagnosis.