Hybrid deep learning approach for brain tumor detection using ResNet and EfficientNet

Identification of brain tumors remains a serious challenge. Traditional investigative techniques rely heavily on manual explanation of MRI scans Machine Learning models help to minimize human error, uncover patterns in imaging data, and support effective treatment planning and prognosis. Despite these benefits, ML faces challenges due to limited and similar datasets, which are vital for developing robust deep learning models. Such limitations can lead to overfitting and poor generalization. Moreover, insufficiency of transparency in deep learning decision-making hinders clinical acceptance. To overcome these difficulties, this paper suggests a fusion of a deep learning approach that ensembles EfficientNet and ResNet models. The ensemble approach adapts to varying MRI protocols and tumor heterogeneity, reducing false positives and missed cases. The performance is assessed using common measures including F1-score, recall, accuracy, and precision.