NeuroGrade

NeuroGrade is a Human & AI collaboration tool that helps radiologists grade brain tumors more accurately by learning from expert disagreements and analyzing all MRI sequences (T1, T2, FLAIR, contrast). Unlike traditional AI based systems, it trains on multi-expert annotations using improves segmentation masks, and transparently shows how other doctors graded similar cases. The workflow is simple - upload MRI scan onto platform, using it as input into the deep learning segmentation model. Multiple doctors and experts will then assess the segmented scans and input their more detailed analysis. Our multi-label, multi-class, multi-annotator framework will then generate a final verdict based on various factors like majority votes, confidence score as well as reliability score. This reduces grading disagreements from 30–40% to under 15%, saves up to 30% decision time, increases overall diagnostic confidence while reducing diagnoses time from 40-60 minutes to less than 5 minutes.

Our team is building upon prior research conducted in collaboration with a local hospital and university on brain tumor imaging and analysis. This earlier work involved understanding MRI scan characteristics, collecting anonymized imaging datasets, and reviewing current clinical workflows for tumor detection