In developmentDictation to report
In development. Radiology dictation is transcribed with Turkish medical speech recognition and placed into a structured MRRT report. The transcript is stored unchanged.
This product is in development. The timing and accuracy figures on this page are our acceptance targets; the product is not released until they are measured and met.
Turkish medical speech recognition
Structured report from an MRRT template
Transcript stored unchanged; report stays linked to it
Pick the modality and study (e.g. chest CT); an MRRT-compatible template loads. No patient name is needed, the internal accession number is enough.
In live mode the text streams while you speak; or upload an audio file (up to 30 min).
Findings fill the template fields, with impression and ICD-10 code suggestions. Every field is linked to its source sentence in the transcript.
The radiologist edits and signs the draft. The signed report is locked and exported as PDF, plain text or FHIR.
A fictional chest CT dictation
Transcript
Contrast-enhanced chest CT.
An 8 millimetre nodule is seen in the right lower lobe.
No mediastinal lymph node enlargement.
Report draft
Targets: live text on screen within 800 ms of speech (p95); measurements and laterality 100% consistent with the transcript; a draft within 60 seconds for 10 minutes of audio.
The report never edits the transcript; a correction opens a separate version. Findings that are not in the transcript do not enter the report.
Every suggested code exists in the reference table. Nothing is exported before a physician approves it.
A signed report cannot be edited; a correction opens an addendum.
Audio is deleted after 30 days by default; the institution sets the period. The speech recognition key never reaches the browser.