Komparasi Deep Learning Convolutional Neural Network (CNN) dan Deep Learning CNN- Support Vector Machine (SVM) untuk Identifikasi Tumor Otak dan Payudara

Siti Nur Khalisha (Faculty of Science and Mathematics, Diponegoro University, Indonesia)
Pandji Triadyaksa (Faculty of Science and Mathematics, Diponegoro University, Indonesia)
Ngurah Ayu Ketut Umiati (Faculty of Science and Mathematics, Diponegoro University, Indonesia)

Jurnal Fisika Unand
Vol 15 No 4 (2026)

The identification of brain and breast tumor MRI images is an important aspect in the development of an accurate and reliable computer-aided diagnosis (CAD) system. This study compares the performance of a single CNN model based on Res-Net 50 and a hybrid CNN-SVM model in classifying different type of tumor from MRI images. The research methods include acquiring brain and breast tumor MRI image dataset from the online repository Kaggle and processing it through image preprocessing steps such as resizing, converting grayscale images to RGB, and performing data augmentation on the training data. In the single CNN approach, Res-Net-50 is used as an end-to-end classifier, while in the hybrid CNN-SVM model, features are extracted from the global pooling layer and classified using SVM. Performance evaluation is carried out using a confusion matrix and chart comparing performance metrics. The research results show that both models achieved 99,16 % accuracy in multi-class brain tumor classification and 98,02 % accuracy in binary breast tumor classification. The CNN-SVM model demonstrated more stable performance across all performance metrics.