A novel approach for the detection of brain tumor and its classification via end-to-end vision transformer - CNN architecture.

The diagnosis and treatment of brain tumors can be challenging. They are a main cause of central nervous system disorder and uncontrolled proliferation. Early detection is also very important to ensure that the intervention is successful and delayed diagnosis is a significant factor contributing to lower survival rates for specific types. This is because the doctors lack the necessary experience and expertise to carry out this procedure. Classification systems are required for the detection of brain tumor and Histopathology is a vital part of brain tumor diagnosis. Despite the numerous automated tools that have been used in this field, surgeons still need to manually generate annotations for the areas of interest in the images. The report presents a vision transformer that can analyze brain tumors utilizing the Convolution Neural Network framework. The study's goal is to create an image that can distinguish malignant tumors in the brain. The experiments are performed on a dataset of 4,855 image featuring various tumor classes. This model is able to achieve a 99.64% accuracy. It has a 95% confidence interval and a 99.42% accuracy rate. The proposed method is more accurate than current computer vision techniques which only aim to achieve an accuracy range between 95% and 98%. The results of our study indicate that the use of the ViT model could lead to better treatment and diagnosis of brain tumors. The models performance is evaluated according to various criteria, such as sensitivity, precision, recall, and specificity. The suggested technique demonstrated superior results over current methods. The research results reinforced the utilization of the ViT model for identifying brain tumors. The information it offers will serve as a basis for further research on this area.