Skin Lesion Detection Using Deep Learning

Skin lesion can be deadliest if not detected early. Early detection of skin save many lives. Artificial Intelligence and Machine learning is helping healthcare in ways so the diagnosis lesion. Computer aided help clinicians detecting cancer. The study was conducted to classify seven classes using very powerful convolutional neural networks. two pre trained models i.e DenseNet Incepton-v3 were employed train model accuracy, precision, recall, f1score ROC-AUC calculated for every class prediction. Moreover, gradient activation maps also used aid determining what are regions image that influence make a certain decision. These visualizations explain ability model. Experiments showed performed better then Inception V3. Also it noted highlighted different predicting same class. main contribution introduce medical classification will understanding decisions It enhance reliability Also, optimizers with both compare accuracies.