Temporal brain tumor progression tracking using deep learning and 3D MRI volume analysis

Cancer is among the most prevalent diseases globally. Concurrently, advances in artificial intelligence are revolutionizing brain tumor diagnosis by offering greater consistency and improved accuracy. This research introduces NeuroSight, a novel technique for tracking brain tumor progression. NeuroSight employs advanced deep learning algorithms to detect, segment, and quantify brain tumor volume from 3D MRI scans. The method combines multi-class EfficientNet and AGSE-VNet models for precise tumor detection and segmentation. NeuroSight’s progression tracker subsequently measures and contrasts the segmented tumor regions over time for individual patients. The efficacy of NeuroSight was tested through a series of experiments. The obtained results show 31.3% improvements in the F1-score using EfficientNet. Also, the AGSE-VNet component achieves 2% and 155.8% better Dice Coefficient and Hausdorff Distance than alternative methods.