Your findings.
A report in your style.

An editable structured draft based on the findings and context you provide, using your style profile if you have set one up. Review all wording, including any technique, comparison or impression, against the actual study before use.

No. You can draft from dictated or typed findings alone. If you attach images, remove patient identifiers from visible content first. The browser strips EXIF and DICOM metadata, but that does not anonymise text printed within an image.

You paste two or three of your own reports into the style wizard. It extracts how you write: bullets or narrative, compact or spacious, whether you include technique and comparison, impression or conclusion or both, precise or approximate measurements, tone, whether you make recommendations, and drafts that way from then on. You can preview the result, change any setting by hand, or delete the profile entirely. It reads samples you choose to paste; it does not watch your work.

Yes, two ways. In Chrome or Edge, connect the device itself over USB: press any button on it once and that button is bound, the same native device integration an enterprise dictation system uses. Or bind a keystroke through your mic driver instead. Hold-to-record or tap-to-toggle, and it works while the cursor is in the findings box. If you also run PowerScribe or SpeechExec on the same desk, bind a button it ignores (F1 to F4, INS/OVR, EOL) rather than the record slide, and ReportsRad only reacts while its own tab is focused, so the two never fight over the mic. Worth knowing: this has been verified in code and in the browser, but nobody has yet pressed a real SpeechMike against it.

Transcription runs on Deepgram Nova-3 Medical, a speech model built for medical dictation rather than a general-purpose one, with a correction pass for common mishears on top. We publish no accuracy percentage, because we have not measured one on your voice and your caseload; the 7-day trial is the honest test. If you are a standalone or independent practice without an enterprise dictation system, this is the part built with you in mind: a medical-vocabulary engine and your own SpeechMike or pedal, in a browser, with no installation, no contract and no per-seat minimum.

No. ReportsRad has no diagnostic or treatment intended purpose, carries no UKCA or CE mark, holds no FDA clearance and is not registered with the MHRA. It writes report wording and prompts checks. Section 8 of the Terms sets out the regulatory position and where responsibility sits: with you, on every report you sign.

No. ReportsRad does not connect to any PACS or RIS, on any plan. You upload anonymised images or screenshots yourself, or paste from the clipboard, and you copy the finished report out. If PACS integration matters to you, say so by email; it does not exist today and we will not pretend otherwise.

Drafting is not restricted by modality: you pick the modality and body part and it drafts accordingly. What is limited is the structure checklist, which exists for seven study types: CT chest, CT abdomen, MR brain, MR knee, MR shoulder, MR spine and US thyroid. Decision-timer reference values cover 19 study types across CT, MR, US and radiographs.

We have not measured a time saving. Try it with de-identified examples from your usual study types and judge the time needed to draft, check and transfer the report.

ReportsRad brings dictation, a saved reporting style, report editing and review prompts into one radiology reporting workspace. Try the workflow with de-identified examples to decide whether it fits the way you report.

A CAD system marks findings on images and is regulated as a medical device for that purpose. ReportsRad does neither. It writes words around findings you have already made. Nothing it says about an image should be treated as a detection, and it is not an alternative to a device cleared for one.

Uploaded images are stored in Cloudflare R2 and deleted within 24 hours by a daily job. Studies and drafts are stored under your account for session restoration. Saving to the Library is a separate action; unpinned library reports are deleted 12 months after their last edit. Remove identifiers before entering information or uploading images. See the privacy policy for processing details and use Settings to export your data or delete your account.

There is no BAA in place today, and ReportsRad is not designed to receive PHI. Do not enter patient identifiers. Remove identifiers from text and visible image content before use; metadata stripping alone is not enough. Check your institution’s requirements and contact support@reportsrad.org if a BAA is required.

Unless you cancel before the trial ends, your Pro subscription simply continues at the plan you picked. Cancel in time and you pay nothing: your account moves to the Free plan, 5 report drafts a day, every feature still there, and your reports and history stay where they are. There is also a 7-day refund on any payment.

Individually, yes: trainees can use it like anyone else. There is no group, cohort or supervisor view in the product: nobody can see another user's data, including programme directors. If departmental use interests you, email support@reportsrad.org and it would be arranged by hand.

A UK CCT trained radiologist, through Aldaim Digital Ltd, registered in England and Wales (company 17327688) and based in Oxford. Contact support@reportsrad.org with questions or feedback.

A full refund within 7 days of any payment: email support@reportsrad.org. After that you can cancel at any time and keep access until the end of the period you have paid for. The Refund Policy has the detail.