Optimasi Deteksi Kanker Kulit Dalam Meningkatkan Keakuratan Dengan Penerapan Efisien Convolutional Neural Network Dan Model EfficientNetB2

Abstract

Skin cancer, associated with ultraviolet exposure, can result from genetic mutations in skin cells. Risk factors include family history, fair skin, moles, weakened immunity, and solar keratosis. Diagnosis involves skin examination and biopsy. Previous studies using Convolutional Neural Networks (CNNs) successfully classified skin cancer with up to 99% accuracy. This method is effective in detecting and classifying skin diseases. The image classification model development process involves structured steps. The dataset is divided into training, validation, and testing sets. Data augmentation is performed with ImageDataGenerator to enrich the dataset. The CNN model (EfficientNetB2) is customized and trained for 50 epochs. Evaluation of the test data includes metrics such as loss, accuracy, precision, recall, and F1- score. Visualization of Classification Reports and Confusion Matrix ensures in-depth analysis of model performance and focuses attention on predictions. The study uses Kaggle's "HAM10000 Preprocessed Data" with 11,644 data and three attributes, showing variations before normalization. The CNN model peaks at the 8th epoch with 86% accuracy, but there is a risk of overfitting. Evaluation using Classification Report and Confusion Matrix provides detailed insight into the model's performance on each skin cancer class, supporting diagnosis and management. This article highlights the positive impact of using the EfficientNetB2 model in skin cancer detection
through an efficient Convolutional Neural Network (CNN). The model's
optimal size and structure produce superior feature representations. Evaluation, including Classification Report and Confusion Matrix, demonstrates the model's superiority in classifying skin cancer types, especially 'akiec' and 'bcc', with high accuracy, significantly improving detection performance.