SKIN CANCER DETECTION BY USING CONVOLUTIONAL NEURAL NETOWORK

• Main Author: William
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/82807
• Institution: Institut Teknologi Bandung

Skin cancer is one of the diseases with increasing cases over time. Skin cancer is caused by DNA mutations in cells within the body. Although skin cancer can be visible to the naked eye, many people cannot distinguish between skin cancer and other skin conditions. In the medical field, doctors perform physical examinations to check for skin cancer. An alternative method doctors use to examine cancerous cells is biopsy. However, biopsies take considerable time and doctors can sometimes misdiagnose. Therefore, there is a need for another method to improve the speed and accuracy of doctors in diagnosing patients. In this study, a deep learning-based model was developed to enhance the speed and accuracy of doctors in diagnosing skin cancer. The model used a convolutional neural network (CNN), which is designed to process and analyze visual data such as images. The types of models in this study can be divided into three categories: self-designed model, Residual Network (ResNet), and Densely Connected Convolutional Networks (DenseNet). In model development, parameters and architectures were varied to find the best model and assess the impact of parameters on the model. The research results indicate that the use of parameters influences the resulting model. Among the models developed, DenseNet169-F with a batch size of 16 was selected as the best model for classifying skin cancer.