Hi, I'm Shashwat Jain

Brain Tumors are among the most complicated and serious neurological disorders due to their unpredictable structure, uncontrollable growth patterns of brain cells and several subtypes - over 120 classified by World Health Organization. Prior and accurate detection is critical for immediate treatment and significantly influences patient prognosis and survival rates. However, manual examination using medical imaging is time consuming, subject to human error, and inconsistent, potentially resulting in delayed or incorrect assessments. This paper explores advanced automated approaches for brain tumor classification that leverage recent developments in transformer based DeiT models to enhance diagnostic accuracy and reliability. We demonstrate a novel approach using Vision Transformers, specifically a DeiT (Data-efficient Image Transformers) architecture for classifying the types of brain tumors. Our model depends on the deit-small-patch16-224 variant, which has been pre-trained on ImageNet and fine-tuned for tumor classification tasks.