Brain Tumor Classification with ResNeXt and its Comprehensive Evaluation

Brain tumors are a significant health risk and require prompt diagnosis and effective treatment. Overfitting, poor accuracy, high computational demands, and time-consuming optimization steps are frequent occurrences with conventional imaging methods. The purpose of this study is to investigate how to apply the ResNeXt deep learning (DL) algorithm for automated brain tumor detection during magnetic resonance imaging (MRI) scanning. The model's robustness was enhanced by using a dataset of annotated MRI images. Compared to traditional methods, the ResNeXt model, which is integrated with RCNN feature extraction, exhibited superior sensitivity and specificity in identifying distinct tumor types. The model's tumor classification accuracy and overall effectiveness are enhanced by using the split-transform-merge approach. The research indicates that ResNeXt has a higher accuracy and classification performance than conventional convolutional methods. The proposed method achieves a brain tumor classification accuracy of 98% and contributes to improving diagnostic precision in neuro-oncology.