Brain Tumor Detection and Classification Using Fine-Tuned CNN with ResNet50 and EfficientNet

Muhammad Ali Sultan (Universitas Mercu Buana)
Christopher Marco Angelo (Universitas Mercu Buana)
Muhammad Alkam Alfariz (Universitas Mercu Buana)
Dinda Fatimah Kautsarina (Universitas Mercu Buana)
Dhani Amanda Putra (Universitas Mercu Buana)
Muhammad Sharjil Ashfaq (Universitas Mercu Buana)
Hadi Santoso (Universitas Mercu Buana)
Genoveva Ferreira Sores (Universitas Mercu Buana)

International Journal of Informatics and Computation
Vol. 6 No. 1 (2024): International Journal of Informatics and Computation

Brain tumors have become a leading cause of mortality worldwide. Detecting and classifying brain tumors accurately at the initial stages is challenging due to their complex and varying structure. In this study, an improved fine-tuned model based on Convolutional Neural Networks (CNN) with ResNet50 and U-Net is proposed. The model works on the publicly available TCGA-LGG and TCIA dataset, which consists of 120 patients. The fine- tuned ResNet50 model outperforms the CNN model in brain tumor classification and detection using MRI images. Accurate and timely diagnosis of brain tumors is critical for successful treatment of the disease. Early detection not only aids in the development of better medication, but it can also save a life in the long run. The domain of brain tumor analysis has efficiently applied medical image processing ideas, particularly on MR images. This paper presents segmentation using Convolutional Neural Networks (CNN) architecture with ResNet50 and EfficientNet as backbones.