Data Complexity Based Evaluation of the Model Dependence of Brain MRI Images for Classification of Brain Tumor and Alzheimer’s Disease

The convolutional neural networks (CNN) have shown promising results for various classification problems over the past years. However, selecting CNN architectures is still challenging as each architecture performs differently with same dataset. This research aims to evaluate dependence of brain MRI on predictive models based complexity data Brain Tumor and Alzheimer’s Disease. Our proposed approach has three parts. First part pre-processing which mainly focuses class balancing estimation complexity. second uses stratified k-fold cross-validation more reliable results. last corresponds implementation four applying described methods. paper compares performance rigorous experimentation variants namely S-CNN (CNN trained from scratch), ResNet50, InceptionV3, Xception two image datasets evaluated without use Principal Component Analysis (PCA). work benchmarks by comparing average scores Accuracy, Precision, Recall, F1 score, AUC score five-fold cross-validation.