Brain Tumor Detection in Magnetic Resonance Images Using Swin Transformer

Abstract

The manual detection of brain tumors from MR images is time-consuming and prone to errors, necessitating the adoption of computer-assisted approaches. The role of artificial intelligence (AI) and its subsets, machine learning (ML) and deep learning (DL), is explored in automating brain tumor diagnosis. This paper discusses the severity of brain tumors, emphasizing their prevalence and low survival rates. This research explores the application of the Swin Transformer for the classification of brain tumors. The research presents the effectiveness of Swin Transformer in analyzing MRI images of different classes of brain tumors, including Glioma, Meningioma, Pituitary, and a class with no tumor. The proposed model incorporates image enhancement techniques and data augmentation methods to improve training efficiency. Results indicate that Swin Transformer outperforms other state- of-the-art models, achieving a high validation accuracy of 86.87% in brain tumor detection. The findings highlight the potential of Swin Transformer for small datasets and medical imaging tasks, offering a promising approach to enhance the accuracy and efficiency of brain tumor classification in medical imaging research.