The Impact of Data Augmentation Methods on The Effectiveness of Brain Tumor Classification
Abstract
This scientific article investigates the impact of data augmentation methods on the effectiveness of brain tumor classification using deep learning techniques. Brain tumor diagnosis from MRI images is a critical task in medical imaging, where limited and imbalanced datasets often reduce model performance and generalization ability. Data augmentation is widely used to overcome these limitations by artificially increasing dataset diversity. In this study, various augmentation techniques such as geometric transformations, intensity adjustments, and GAN-based synthetic image generation are analyzed in terms of their influence on classification accuracy. Experimental results show that data augmentation significantly improves model performance, with accuracy gains ranging from 5% to 20% depending on the applied method. The findings confirm that carefully selected augmentation strategies enhance the robustness and reliability of brain tumor classification systems.