Abstract: The World Health Organization (WHO) identifies brain tumors as one of the leading causes of death in the world. This disease is challenging to identify because of its complexity and cunning character. Because of the high risk of clinical occurrences, persistent brain tumors illness is a severe public health issue worldwide. Despite the general consensus that persistent brain tumors disease has considerable interactions with elevated risks of vascular events, end-stage excretory organ disease, and all-cause mortality, there is still inadequate information on individuals. Deep learning (DL), a branch of machine learning, has recently shown impressive results, particularly in tasks like classification and segmentation. Imaging can be done in various ways to look for brain tumors. This paper presents a novel solution to a challenging problem through the development of a Deep Hybrid Transfer Learning Based Convolutional Neural Network, referred to as EDTLM. Our proposed approach commences with the preprocessing of MRI images, followed by the design and evaluation of the EDTLM. Through rigorous assessment of loss and accuracy metrics, we sought to enhance the model's performance. To achieve this, we harnessed the power of deep transfer learning by fusing features extracted from these models, thereby significantly improving the classification accuracy for three distinct types of tumors. Our results, obtained from the application of the Hybrid Deep Transfer Learning model, namely the Effinception Deep Transfer Learning Model, underscore the potency of combining deep learning models. This approach not only enhances accuracy in multiclass classification problems but also addresses the challenge of overfitting in the context of imbalanced datasets. Our proposed model aspires to achieve an impressive classification accuracy of up to 99.77%, setting a new benchmark in this domain. Furthermore, our framework consistently demonstrates its competitiveness when juxtaposed with other state-of-the-art studies. This work not only offers innovative insights but also holds great promise for advancing medical image classification tasks.

Brain Tumor, Analysis, Detection, Classification, Hybrid Learning, Ensemble Learning, Transfer Learning, Convolutional Neural Network.