What You Get
Every scan produces a formal clinical imaging report. Structured like a radiologist's report, exportable to a multi-page PDF, with a live follow-up chat that never loses context.
EXAM TITLE + VIEW
Auto-detected — "CHEST X-RAY · PA View", "CT BRAIN · Axial", "MRI KNEE · Sagittal".
STRUCTURED FINDINGS
6–12 discrete clinical sentences. Each with confidence and a 3×3 region tag.
IMPRESSION
One-sentence diagnostic conclusion. "No abnormality detected" when normal, or the leading diagnosis otherwise.
CLINICAL ADVICE
One-line recommendation — often "clinical correlation recommended" or a specific next step.
DIFFERENTIAL DIAGNOSIS
3–6 conditions ranked most-to-least likely, each with a short rationale.
MANAGEMENT
3–6 specific next steps — imaging, labs, referrals, or conservative management.
PDF · MD · JSON EXPORT
Multi-page PDF with letterhead. Markdown for docs. Raw JSON for pipelines.
FOLLOW-UP CHAT
Ask anything about the report — image-grounded, zoom-aware. $0.01 per turn. Saved forever.
The Follow-Up Chat
A report is a starting point, not an endpoint. After a scan completes you can ask anything about it — and the model will re-open the original image if it needs to look closer.
THE IMAGE IS STILL IN CONTEXT
Every follow-up turn has access to the original scan. If you ask about a specific region, the model crops that quadrant and re-answers with the zoom.
VISION OR TEXT — AUTOMATIC
Image-grounded questions go to the 27B vision head. Definitions, prognosis, and management questions go to the 20B text head. You never choose.
$0.01 PER TURN
Each follow-up turn debits a cent from your wallet. Failed turns refund automatically. No subscription, no minimum.
NOTHING IS EVER LOST
Every message, every image, every report is persisted to your private history. Close the tab and come back a year later — it's all still there.
Four Report Shapes
Same underlying engine, four different report layouts — one for each scan mode. The report card adapts to what the mode produces.
STANDARD REPORT
Findings · Impression · Advice · Differential · Management. Follow-up chat enabled. Optional 2-pass localization verification.
→ FULL CLINICAL REPORTMULTI-MODEL REPORT
Same layout as Standard plus per-finding agreement scores computed across two independent vision passes.
→ WITH AGREEMENT SCORESABNORMALITY REPORT
Bounding boxes drawn on the source image, severity colour-coded (mild · moderate · severe · critical), numbered against findings list.
→ ANNOTATEDCOMPARE REPORT
Interval change: new findings, resolved findings, and stable findings between two studies — grouped by region.
→ PRIOR VS CURRENTSample Report Structure
Below is the shape of a typical structured report — the exact JSON fields the model produces.
Generate one now
Attach a scan, enter patient details, and get a report in 20–40 seconds. Free first scan.