Comparative Study of Automated Brain Tumor Cl...

Comparative Study of Automated Brain Tumor Classification using MR Scans

Author Name : Ritu Shukla, Pramod Kumar Singhal

ABSTRACT The accurate classification of brain tumors is critical for timely diagnosis and treatment planning, yet manual radi-ological assessment remains prone to subjectivity and human error. This study explores the application of machine learning (ML) techniques to automate and enhance the differentiation of brain tumor types using medical imaging data. Magnetic resonance imaging (MRI) scans are preprocessed through noise reduction and intensity normalization, followed by segmentation to isolate the region of interest. Key morphological and textural features are subsequently fed into multiple classifiers, including support vector machines (SVM), random forests, and k-nearest neighbors (KNN). Comparative evaluation on a publicly available dataset demonstrates that the ensemble model achieves superior performance, with a classification accuracy exceeding 94% for distinguishing glioma, meningioma, and pituitary tumors. No-tably, the random forest classifier exhibits robust generalization, minimizing overfitting while maintaining high sensitivity and specificity. These findings underscore the potential of ML-based systems as reliable decision-support tools in clinical neuro oncology, reducing diagnostic turnaround time and inter-observer variability. Future work will focus on deep learning integration for end-to-end feature learning and validation on larger multi-center datasets to ensure clinical translatability