Machine learning · Local deployment · No external APIs

Clinical-grade clarity for brain MRI review workflows

Upload MRI slices, run an on-device ensemble model, and export traceable PDF reports — built for minor projects that still feel like production software.

Hashed passwords · Role-based access · SQLite audit trail

Live pipeline Online
  1. Secure signup & session
  2. Validated JPG/PNG upload with preview
  3. HOG + RandomForest inference (re-trainable)
  4. History, PDF/text export, dashboards
  5. Admin training, metrics, and global reports

Everything your submission checklist asks for

Authentication, uploads, multiclass prediction with confidence, downloadable artifacts, and a full admin control plane — all in Flask + SQLite.

Accounts & recovery

Signup with duplicate checks, login, hashed passwords, security-question reset, and clean logout redirects.

MRI intake

Type and size validation, local disk storage, preview & replace before inference, timestamped results.

Trainable ML core

Train/retrain from labeled folders, persist joblib model, confusion matrix + classification report in admin.

History & export

Searchable history, per-result detail, PDF/text downloads, user deletions, and dashboard rollups.

Admin cockpit

Separate admin session, user lifecycle, dataset browser, training, reports, feedback inbox, settings.

Reporting

User-wise, date-wise, and overall HTML exports generated entirely inside the project — no cloud APIs.

Ready to demo your pipeline?

Spin up the seed database, sign in as a sample user, and walk examiners through upload → predict → PDF in one sitting.