Polishing & Proofreading

How RadPal polishes and proofreads your dictated findings before report generation.

What It Does

Before your report is generated, RadPal automatically proofreads and polishes your dictated findings. This step cleans up dictation artifacts, grammar issues, and formatting inconsistencies while preserving your clinical meaning. The polished text is what the AI uses to build your final report.

Polishing corrects:

Spelling and transcription errors — typos and misrecognized words from dictation are corrected.
Grammar and punctuation — sentence structure, punctuation, and capitalization are cleaned up.
Phrasing and flow — text is adjusted to read naturally in a concise, professional radiology style.

No medical information is added, removed, or inferred. The polishing step only changes form, not content.

Polishing vs. Transcription Corrections

RadPal has two separate mechanisms for cleaning up dictated text, and it helps to keep them apart:

Transcription Corrections — a deterministic find-and-replace list you define yourself in Settings → Advanced → Transcription Corrections. Each entry maps a commonly mistranscribed word to its correct term (e.g., "aortobiliac" → "aortoiliac"). Matching is whole-word and case-insensitive, with the original capitalization preserved. Replacements are applied both to your dictated findings before they reach the AI and to the generated report that comes back. The same word is corrected the same way every time — no AI involved.
AI polishing — the LLM proofreading pass described on this page. It handles grammar, punctuation, and phrasing, and flags ambiguous words for your review rather than guessing.

If your dictation software consistently mishears a specific term, add a Transcription Correction for it — that guarantees the fix rather than relying on the AI to catch it.