An Advanced Brain Tumour Detection Framework Using Improved Fast Mask R-CNN Techniques for Enhanced Accuracy

To enhance the precision of diagnosis, this research provides a new structure for identifying brain tumors that integrates an Improved Fast Mask Region based Convolutional Neural Network (IFMRCNN) with complex image processing algorithms. Identification of brain tumors is still a significant difficulty in medical imaging since successful therapy depends on early and accurate identification. By precisely identifying tumor boundaries in brain MRI scans, the proposed model makes use of an enhanced version of FMRCNN is intended to enhance separating and categorizing performance. Sophisticated image processing methods are used to improve the extraction of features, allowing the model to more accurately differentiate between tumor and non-tumor areas. By lowering false positives and true negatives, the structure seeks to maintain computing power while greatly increasing detection rates. The proposed strategy outperforms conventional techniques in terms of reliability, specificity, and sensitivity, according to the results of experiments, which makes it a viable tool for support with clinical choices as well as initial diagnosis in neuroimaging. IFRCNN was applied using the Open CV package in conjunction with an SVM algorithm. Using 218 photos as an initial training set, this study's approach demonstrated a 100% success rate in classifying meningiomas and 87.5<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> accuracy in classifying gliomas, with an average conviction score of 94.6% in the segments of meningioma tumors. A neurologist does a manual examination and ground truth assessment to verify the divided tumor areas