3 November 2026

Radiology reports: harmonising quality without standardising practices

News

How can a network deliver consistent radiology reports while respecting each physician's habits? At JFR 2026, Doctreen and Réseau Vidi compared their experience around one conviction: artificial intelligence should bring the quality of reports into line, without imposing a single way of working.

Different practices, the same medical standard

Free dictation, speech recognition, template or structured reports: within the same radiology group, working methods can vary considerably.

This diversity reflects each practitioner's experience and preferences. Yet it raises a challenge: how can referring physicians be guaranteed clear, complete and immediately usable information, whichever radiologist performs the examination?

The question matters especially in oncology, where comparing successive examinations requires precise information, consistent measurements and easily identifiable conclusions.

At the symposium held at JFR 2026, Doctreen and Réseau Vidi explored a different approach to standardisation.

The goal is not to impose a single report template, but to guarantee a shared foundation of medical quality.

Standardise the result, not the radiologist

Imposing the same writing method on every radiologist can create additional constraints, slow down their practice and meet resistance.

The alternative is to preserve everyone's freedom while harmonising the essential information in the report.

Three priorities emerged from the discussions:

  • Medical quality: ensuring complete, structured reports and concise conclusions.
  • Practitioner freedom: keeping one's own dictation habits, vocabulary and organisation.
  • Intelligent assistance: making it easier to include medical classifications and guidelines without adding steps to the workflow.

This approach makes harmonisation possible across a whole network, without erasing each radiologist's individual style.

AI in support of medical reasoning

For Michel Benisty, co-founder of Doctreen, artificial intelligence should not simply automate report writing.

Its real potential lies in its ability to draw on medical knowledge and support the practitioner's reasoning.

This is precisely the approach developed by Doctreen, which combines generative AI, medical knowledge graphs and decision trees built from radiologists' expertise.

The physician keeps dictating naturally while the tool helps structure the information, include the relevant classifications and flag the points that need particular attention.

This vision also opens up new perspectives: gradually entrusting AI with certain detection, measurement and report preparation tasks, so that radiologists can refocus on interpretation and medical decision-making.

The aim: let technology carry part of the production burden, without taking control of the diagnosis away from the physician.

A shared vision to discover in our podcast

How can reports be harmonised across a network? How can medical guidelines be built in without weighing down practice? How far can AI support the radiologist?

Through clinical feedback and a presentation of Doctreen's technological vision, this symposium sheds concrete light on how radiology practice is evolving.

Listen to the full Doctreen × Réseau Vidi discussion in our podcast recorded at JFR 2026.

Discover Doctreen Spark

Turn your free dictation into a structured report, while keeping your working habits and full control over your conclusions.

Radiology reports: harmonising quality without standardising practices — Doctreen