A Novel Approach To Brain Tumor Detection using AI ML, DL, with Synthetic Data Augmentation

Effective treatment and better patient outcomes depend on early brain tumor identification. Even if they are trustworthy, traditional diagnostic techniques frequently need manual analysis, which can be laborious and arbitrary. In order to improve brain tumor diagnosis from MRI scans, this work introduces a unique method that combines Support Vector Machine (SVM) clustering, Statistical Gray Level Co-occurrence Matrix (LBP) features, ResNet50, and synthetic data augmentation. SVM clustering overcomes the drawbacks of linear clustering techniques by accurately segmenting tumor areas. Texture-based characteristics like homogeneity and contrast are extracted using LBP, offering vital information for categorization. By utilizing its deep learning architecture, the ResNet50 model attains great accuracy after being adjusted based on the retrieved characteristics. By mitigating data scarcity, synthetic data augmentation ensures the resilience of the model. Improved accuracy, sensitivity, and specificity are demonstrated by experimental findings on publicly available datasets, indicating the efficacy of this strategy in providing scalable, dependable solutions for automated brain tumor diagnosis.