TRANSFORMING BRAIN TUMOR DIAGNOSIS: VISION TRANSFORMERS COMBINED WITH ENSEMBLE TECHNIQUES

Abstract

Brain Tumor (BT) is widely recognized as one of the most prevalent illnesses worldwide, affecting approximately 24,810 people in the year 2023. Most people suffering from brain tumor disease belong to the Southeast Asian and Western Pacific regions. Medical diagnostics using artificial intelligence and deep learning models demonstrate efficacy in addressing critical health challenges in initial disease and detection of intervention of BT. In this paper, we proposed ViT along with ensemble learning models for multiclass brain tumor classification and detection. The proposed work aims to provide the novel best solution to the problem of brain tumor detection using a deep learning approach. Ensemble Learning obtained 96% accuracy and loss of 0.13 with an F1-score, precision, and recall of 0.96. The comparative result shows that Vision Transformer ViT obtained 90% accuracy and loss of 0.30 with an F1-score, precision, and recall of 0.89 on the brain tumor MRI dataset containing 7023 images, which is further divided into train and test. The promising results showcase the potential of this proposed system in early and accurate brain tumor detection. The proposed system can be used in the early detection of brain tumors.