Ensemble XMOB Approach for Brain Tumor Detection Based on Feature Extraction

Brain tumors are a serious health threat in adults. These fast-growing abnormal cell masses disrupt normal brain function. Doctors use various imaging techniques to identify the specific type, size, and location of brain tumors in patients. Accurately identifying and classifying brain tumors is crucial for understanding how they develop and progress. Magnetic Resonance Imaging (MRI), a well-established medical imaging technique, plays a vital role in this process by assisting radiologists in investigating the location of the tumor. Previous models frequently encounter a compromise between accuracy and computational efficiency, lacking an approach that successfully integrates both aspects.This study introduces an innovative ensemble model termed as “XMob Approach” that combines the deep features extraction abilities of Xception with computational efficiency of MobleNet for binary classification of brain Tumor. The Xmob Approach leverages the strengths of both architectures : Xception depthwise seperable convolutions allow for detailed feature extraction whereas MobileNet’s lightweight structure ensures efficient computation making it suitable for real life application. This combination aims to enhance in medical diagnostics, promising enhanced accuracy and efficiency. This study explores the potential of integrating these pre-trained architectures to provide real-time, automated diagnostic assistance, improving the speed and precision of brain tumor detection. In our methodology pre-processed MRI scans undergo feature extraction through Xception model, capturing complicated patterns indicative of tumor presence. Simultaneously MobileNet processed these images emphasizing computational efficiency without compromising on performance.The output of both the modesl are then integrated using ensemble technique to improve overall classification accuracy. By integrating the complementary strengths of Xception and MobileNet , the XMob Approach represent a significant step towards the field of medical diagnostic promising improved outcomes for patients through advanced technology. DOI: https://doi.org/10.52783/tjjpt.v45.i03.7253