An Experimental Brain MRI Research Project
Built by a high school student, this free brain scan analyzer explores how deep learning models behave in development and on real uploaded scans. The models frequently fail on real uploads. Their output is for education and research, not diagnosis.
Read why real-world results are lower than development metrics.
Explore the Models
- Original Detection - RF-DETR (Medium), trained on DWI, draws boxes around regions it flags as possible tumor
- Classification Models - YOLO26 models draw boxes or pixel outlines and predict one of three tumor types
- No Account Required - No registration, no sign-up, no personal information needed
- Multiple Format Support - Upload DICOM, PNG, and JPEG images; review model output and PDF reports
Research & Learning
- About the Project - why a high school student started this research project and what real uploads have shown
- Learn: MRI & DWI - how MRI and diffusion-weighted imaging work, and why AI research in medical imaging is difficult
- Model Transparency - architectures, training data, development metrics and why results on real uploads are lower
- History of Updates - a dated record of model versions and application changes