Skin Cancer Detection Optimal Deep Neural Network-Driven Computer-Aided Diagnosis Model

Abstract

Nowadays,lots of persons are suffering from numerous types of skin cancer particularly in America. Recently, imaging-based Computer Aided Diagnosis (CAD) approach is extensively used to monitor and identify the skin cancer. In this paper, an automatic model was intended to categorize the various skin disease melanoma, nevus pigmentosus, dermatofibroma, and squamous cell carcinoma using CNN.The model consists of a fully connected layer, a fully connected layer based on 3 × 3 filter sizes, and softmax activation. This model employs SGD, RMSprop, Adam, and Nadam optimizer.The results show the Adam optimizer achieves with 0.0255 loss and accuracy is 99% performance than other optimization techniques with CNN model to classify the skin cancer types. Using theexecutionoutcomes, the modeldepicts that the proposed approach is hopeful to use as atraditional tool for medical personnel.