Suggested classification,
with the source at hand
GUIDE suggests a category for physician review based on the findings provided. RadCommons provides the structured, cited criterion; the physician checks that it is current and decides.
What GUIDE covers
- SUGGESTION AT THE RIGHT POINT · YOUR DECISIONBI-RADS 2BI-RADS 4A ✓TI-RADS 3LI-RADS…SYSTEMS
The classifications of daily practice
BI-RADS, LI-RADS, PI-RADS, TI-RADS, Lung-RADS, O-RADS and the other systems in the RadCommons corpus, each in its current version.
- GROUNDING
Cited criterion, not invented
Each suggestion points to the criterion that supports it, with a citation and locator. The AI shows where it came from, and the physician decides whether to agree.
- REVIEW
A suggestion, never a decision
The category enters the report as a proposal for physician review. The final word always belongs to the radiologist who signs.
The classification appears with its criterion, not as a guess.
GUIDE is a layer of the clinical workflow, connected to the exam context, to physician review and to what happens after signature.
Human supervision,
by architecture.
GUIDE makes the grounding visible. The suggestion can be accepted, adjusted or rejected; the final decision remains the radiologist's.
Less operation.
More image.
Memory and loose tabs create variation. Versioned, cited criteria reduce friction without turning a guideline into autopilot.
“A visible source lets you check the rule; even so, context and the final decision remain medical.”
Why grounding matters
A wrong classification is not a small error: it changes management, changes follow-up, changes the patient's life. A system that outputs a category without showing the criterion asks for blind trust, and blind trust does not pass an audit and should not pass your review either.
GUIDE works on top of RadCommons, the decision corpus that structures, versions and cites radiology's classification systems. When the suggestion appears, the criterion that supports it is one click away, in the current version of the guideline.
Specification
- FOUNDATIONRadCommons: versioned, cited decision corpus
- OUTPUTSuggested category with the supporting criterion
- UPDATESSystems versioned according to the current guideline
- AUTHORSHIPThe radiologist reviews, edits and signs. Always.
- AUDITSuggestion and decision recorded in the exam log
Direct questions
What happens when the guideline changes?
RadCommons versions the systems. GUIDE then suggests based on the current version, and the log records which version grounded each report, which protects the institution in retrospective reviews.
Can I disagree with the suggestion?
You should, whenever your judgment says so. The suggestion exists to speed up the typical case and standardize the team; the decision is medical and is recorded as yours.
Which systems are covered?
The main ones in daily practice (BI-RADS, LI-RADS, PI-RADS, TI-RADS, Lung-RADS, O-RADS, CAD-RADS, Bosniak, Fleischner and others) and the full RadCommons catalog, which grows with curation by radiologists.
A cited criterion is worth more than a memorized one
BI-RADS, LI-RADS, PI-RADS, TI-RADS and the other systems of daily practice change versions, and on-call memory does not keep versions. GUIDE suggests the category at the right point in the report with the RadCommons criterion alongside: structured, cited and versioned. The decision remains the radiologist's; the difference is deciding with the current source one click away instead of a table open in another window.
For the department, the compound effect is agreement: when the criteria behind each category are recorded, reports from different radiologists become comparable, and a discussion of disagreement becomes a conversation about criteria, not about memory.
What happens after the suggestion
How to handle agreement and disagreement
A suggestion is only useful when the radiologist can see the criterion and accept, adjust or reject it. GUIDE keeps the category as an editable proposal and links the decision to the system and version used on that exam. Disagreeing does not break the flow: the physician's decision goes into the report, and the difference between suggestion and decision remains available for quality review.
This allows a better clinical conversation. Instead of measuring blind adherence to the AI, the team can review where there was disagreement and whether the problem lay in the described finding, the applicable rule or the suggestion. The goal is to reduce avoidable variation without hiding the uncertainty that belongs to the case.
Coverage must be verified, not assumed
BI-RADS, LI-RADS, PI-RADS, TI-RADS and other systems are part of the corpus, but modality, version and use scenario must be confirmed during evaluation. When a system does not cover the case or the finding does not support a category, the safe answer is not to suggest one. The demonstration should include the systems the team actually uses, not just the easiest example in the catalog.
Criterion at hand, physician's decision.
BI-RADS, LI-RADS, PI-RADS, TI-RADS and others: GUIDE suggests the category with the current criterion visible alongside, for physician review. The suggestion saves you looking up the table; the classification remains the radiologist's.
See GUIDE suggesting on your case
Live demo with your exam types and your institution's template, right in the product.