A Deep Learning Framework for Early Brain Tumour Detection Using Hybrid Attention Efficient U-Net Model

Early detection of brain tumours enhances patient survival rates and advances medical practices. Manual detection techniques are labour-intensive and complex. Magnetic Resonance Imaging is widely employed for tumour detection because of its non-invasive nature and ability to provide detailed imaging without the need for painful or intrusive biopsies. However, manually analyzing magnetic resonance images from various angles is challenging and error-prone. This study aims to develop an advanced deep learning framework for accurate brain tumour classification.This paper proposes a novel Hybrid Attention Efficient U-Net model for accurate and automated classification. The proposed approach begins with data collection, followed by pre-processing to enhance image quality and standardize inputs. Essential image features are then extracted to capture key characteristics. These features are input into the classification phase, where the Hybrid Attention Efficient U-Net model combines the strengths of both EfficientNet and U-Net architectures. The method is then computed on three publicly available datasets, ensuring robust validation.The experimental results demonstrate that the proposed approach provides a reliable and efficient solution for automated brain tumour detection and diagnosis, thereby significantly improving diagnostic accuracy of 99.01% and precision of 98.9%.