NeuroNet: Enhanced CNN with FPN and Attention Mechanisms for Brain Tumor Classification and Segmentation Using Explainable AI (XAI)

Segmentation and classification of brain tumors from MRI scans are important medical diagnostic issues. Early and accurate detection is an important determinant of patient prognosis because it allows for early treatment intervention and optimal treatment planning. Manual segmentation of MRI scans is the premise of customary conclusion of brain tumors and is time consuming and vulnerable to error. In this work, we introduce NeuroNet, a novel profound neural arrange architecture that combines CNNs for classification and division of brain tumors, Feature Pyramid Networks (FPN), and attention. NeuroNet classifies and detects various brain tumors such as glioma, meningioma, pituitary tumors, non-tumor tissues without manual intervention. We also use Explainable AI (XAI) techniques, like Grad-Cam, to display areas of the MRI scans responsible for the prediction made by the model, thus making it easier to be transparent and build trust in decision-making. The model has 96.5% segmentation accuracy. With the application of multi-scale feature extraction and attention mechanism, NeuroNet accommodates tumors of varying sizes with better segmentation accuracy. Work of XAI procedures not as it were upgrades the interpretability of the demonstrate but moreover helps in clinicians' comprehension of the rationale behind the model's predictions. Findings demonstrate the guarantee of NeuroNet as an interpretable, autonomous, and dependable framework to apply for brain tumor detection as well as classification in a clinical setting.