Abdullah, FasihJamil, AkhtarAlazawi, Esraa MohammedHameed, Alaa Ali2025-04-182025-04-182024Abdullah, F., Jamil, A., Alazawi, E. M., & Hameed, A. A. (2024, April). Exploring Deep Learning-based Approaches for Brain Tumor Diagnosis from MRI Images. In 2024 IEEE 3rd International Conference on Computing and Machine Intelligence (ICMI) (pp. 1-11). IEEE.979-835037297-7https://hdl.handle.net/20.500.12713/6324Brain tumors significantly impair health due to their aggressive growth, necessitating rapid and accurate detection and classification for effective treatment. Traditional methods relying on medical professionals' manual MRI analysis are time-consuming and resource-intensive. This study leverages Artificial Intelligence-based deep learning approaches to streamline and enhance the accuracy of brain tumor classification from MRI images. We evaluated five advanced deep learning models: a custom CNN, DeepTumor-Net, VGG-16, ResNet-50, and Xception. These models were applied to the 'Brain tumors 256×256' dataset, classifying brain tumors into four distinct categories: no tumor, Glioma, Meningioma, and Pituitary Tumors. These models were trained and then fine-tuned, which further refined their performance. The models were evaluated using the standard evaluation metrics, including accuracy, precision, recall, F1-Score, specificity, and Cohen's Kappa. The final results showed that high accuracies can be obtained for MRI classification using these deep learning models. Notably, ResNet-50 and VGG-16 stood out with test accuracies of 92.6 % and 92.1 %, respectively, indicating their significant potential in medical imaging analysis. © 2024 IEEE.eninfo:eu-repo/semantics/closedAccessBrain Tumor ClassificationConvolutional Neural NetworksDeep LearningMedical Image AnalysisModel Fine TuningConference Object2-s2.0-8519945936210.1109/ICMI60790.2024.10585851N/A