Brain reads
Diffusion and ADC, haemorrhage, masses and mass effect, white matter disease with Fazekas grade, ventricles and posterior fossa.
Monitraq turns a brain or spine MRI study into a structured, evidence-linked draft read in minutes, so radiologists start from the right slices, the right levels and the right side.
Every finding links to the exact slice it was seen on. Every report says what could not be assessed. Nothing reaches a patient before a radiologist signs off.
Diffusion and ADC, haemorrhage, masses and mass effect, white matter disease with Fazekas grade, ventricles and posterior fossa.
Disc contour, alignment and listhesis, canal and both foramina for every level, with full-resolution central sagittal slices.
Every tile carries orientation markers and slice direction, so left, right and spinal level are read from the image, not guessed.
Only a whitelist of technical metadata reaches the AI. Names, IDs, dates and institutions never leave the server.
Red flags and a suggested priority (routine, priority, urgent) so critical studies reach a radiologist first.
Past studies with radiologist reports become scored test cases, so every change to the AI is checked against real reads.
DICOM zip, NIfTI or DICOM files go straight to private storage in Mumbai.
Headers are stripped, series sorted, localizers dropped, and key slices rendered as labelled montages.
Claude reviews the montages against a region-specific checklist and returns a structured report.
Findings link to tile IDs, so the radiologist confirms or corrects in seconds. The raw scan is deleted.
Severity, confidence and evidence tiles for every finding; per-level tables for the spine; explicit limitations when a sequence is missing. The draft never claims more than the images show.
| Level | Disc | Alignment | Canal |
|---|---|---|---|
| L1-L2 | mild desiccation | normal | none |
| L2-L3 | bulge | normal | none |
| L3-L4 | bulge | normal | mild |
| L4-L5 | protrusion (central) | normal | moderate |
| L5-S1 | severe degeneration | anterolisthesis gr. I | mild |
Evidence: S2-13 · S2-14 · S1-08 | Illustrative example
Illustrative: real scores appear in the super admin console.
Super admins upload past studies with the radiologist's final report. The AI reads blind, a second pass scores it finding by finding, and the lessons drive the next version. Every change is re-tested on the whole reference set before it ships.
Every output is a draft for a qualified radiologist and carries a fixed disclaimer.
Scans are deleted after each read; uploads and results stay in private storage in Mumbai.
Doctors are approved by a super admin and only ever see their own studies.
Limitations, missing sequences and low-confidence calls are stated, never hidden.
Dr. Priti Neel Shah is a radiologist and the founder of Monitraq. She holds an MD in Radiology and previously served as an Assistant Professor of Radiology at KEM.
Through Monitraq she is building a smarter remote patient monitoring platform. Monitraq MRI brings the same idea into the reading room: an AI second reader that drafts brain and spine reports, points to the exact slice behind every finding, and always leaves the final call with the radiologist.
Region-specific checklists, level-by-level spine tables, left and right checked on every slice.
Every finding cites the image it was seen on, so it can be verified in seconds.
A draft for review, never a diagnosis, and honest about what it could not assess.
Request access with your work email. A super admin approves new doctor accounts.