Brain Tumour Detection Using Deep Learning

Brain tumors are among the deadliest diseases affecting humans, often leading to rapid fatality if not identified and treated at an early stage. However, relying solely on visual examination can sometimes result in inaccuracies, and seeking expert assistance may be expensive or inaccessible. In this context, deep learning (DL) has emerged as a transformative approach for detecting tumors in medical images. This work proposes a modified DL-based model leveraging EfficientNet-B0, a variant of the EfficientNet (EN) family, to classify MRI scans as tumorous or non-tumorous. The model incorporates multiple deep layers and employs the Soft Max classifier for image classification. A substantial dataset of MRI images was utilized to train and evaluate themodel, addressing a limitation in prior studies, which often used smaller datasets. Notably, few studies have explored the application of the EfficientNet architecture for brain tumor detection. Our model achieved a detection accuracy of 99.97%, with a precision of 91.63%, an F1-score of 86.94%, and a recall of 85.49%. These results indicate that our model outperforms previous EfficientNet-based approaches in terms of detection accuracy.