An Efficient Deep Learning Technique for Brain Abnormality Detection Using MRI Images

Abstract

Abstract This study focuses on leveraging advanced medical imaging techniques, encompassing X-rays and MRIs, to effectively detect brain anomalies, notably tumors. The conventional manual examination approach is time-intensive and often suboptimal. The study proposes a novel method employing machine learning algorithms to categorize 700 patient images as either "brain" or "non-brain" following meticulous labelling and preprocessing. The binary classification comprises "Normal" and "Abnormal" classes, with model accuracy refined through adjustments and augmented training on expanded datasets. Through comprehensive model evaluation including ANN, CNN, VGG-16, and AlexNet, the VGG-16-based model emerges with the highest accuracy at 94.4%. This research underscores the immense potential of advanced deep learning, ensuring swift and precise brain abnormality detection in medical imaging with significant clinical implications.