XXIX Antioquia Radiology Seminar: how the meeting with the profession went

The XXIX Antioquia Radiology Seminar is over, and at NOVA Imaging we come away with one certainty: those two days were worth every hour of preparation. We met radiologists, technologists, imaging center directors and technology leads from institutions across the country, and we left with a notebook full of conversations that are already changing how we prioritize our work.

Here is how it went, what we showed and what we took home.

Being here is not a detail: it is where we come from

NOVA Imaging is a Colombian company, and an Antioquian one. We build RIS/PACS software from here, for institutions operating under the real conditions of the Colombian health system: tight budgets, growing volumes, services that cannot stop, and data regulation that has to be complied with for real, not on paper.

That is why a seminar like this is not just another trade show on the calendar. It is where the region’s radiology profession sits down to talk about what actually hurts and what actually helps. Being there, with the full team and the product running on screen, is the most honest way we know of accounting for what we are building.

And there is something that only happens in person: seeing a radiologist’s face when they understand, in ten seconds of demo, that a task currently eating half their day could stop eating it.

What we showed: artificial intelligence already in production

The conversation about AI in radiology moved past the promise stage a while ago. What we brought to the seminar is what is running today at real institutions, not a concept video.

AI-assisted structured reporting

This was, without question, the demonstration that stopped the most people in front of the screen.

The system takes what the radiologist has already done on the study — the measurements, the marked findings, the dictation, the template for that exam type — and produces a structured, coherent report in seconds, inside the reading workflow rather than in a separate tool. The physician corrects whatever needs correcting, signs digitally, and the report goes out with version control and ICD-10 coding.

The point we repeated most often at the stand, because it is the one that matters: the clinical judgment and the signature remain the radiologist’s. What disappears is the transcription time, not the professional responsibility.

Report quality control and inconsistency detection

Alongside the assisted report we showed the layer that works quietly behind it:

• AI quality control: compares the signed report against the measurements and findings marked in the study, and flags omissions or discrepancies before delivery. A second pair of eyes before signing.
• Real-time inconsistency detection: while the radiologist writes, it flags a measurement, a laterality or a study type that does not match what is marked on the image. The most visible effect at services already using it is the drop in addenda.

Several of the sharpest questions at the seminar pointed exactly there — at laterality and at mismatches between image and text. It is a problem everyone knows about and few discuss publicly. We were glad it got discussed.

Intelligent case prioritization

The other demonstration that generated long conversations was the intelligent worklist: prioritization by urgency, modality, payer contract and SLA, with work lists customizable per radiologist and shareable across shifts.

For a high-volume center, the difference between a flat work list and one that knows what should be read first is not cosmetic: it is turnaround time, it is meeting service agreements, and it is clinical risk under management. More than one service director stayed to sketch out with us what their own list would look like.

Cloud storage without the surprise invoice

The other big topic in conversation was the one nobody puts on the poster but everybody deals with: what to do with studies at ten, fifteen, twenty years.

We showed how our cloud infrastructure works: compression and deduplication, and above all automatic tiering across fast storage, standard storage and cold archive. Recent studies live where speed is needed; studies nobody consults daily migrate on their own to cheaper tiers, without anyone administering it by hand.

The result that matters to whoever signs the budget: a predictable cost projection over ten years and beyond, with backup and disaster recovery included, and the option to deploy in the cloud, on-premise or hybrid depending on what each institution needs.

We were also clear about what we do not promise: downloading images from the cloud has a cost, and we would rather that be known from day one and modeled into the projection than show up as a surprise in year two.

Interoperability: the question that always comes

And it came, as always: “does this connect with what I already have?”

We showed the answer on screen: HL7 v2 (ADT, ORM, ORU), FHIR R4 over a REST API, DICOM and DICOMweb (WADO, QIDO, STOW), modality worklist and IHE profiles. The RIS operates as the orchestrator: it receives the order from the HIS, delivers the worklist to the modalities, controls the reading stage and returns results to the HIS, to the ERP for billing, and to the patient and referring physician portals.

None of that is an extra: it is the minimum condition for a RIS/PACS to coexist with the technical reality each institution already has installed.

What we took home

Beyond the demonstrations, what makes a space like this valuable is what you carry back to the office:

• Conversations with the people who use the system every day. The best product decisions we have made did not come out of a planning meeting: they came out of a radiologist explaining to us, with justified impatience, why something was not working for them.
• A stronger professional network. Reconnecting with customers, colleagues and partners — and meeting the people just discovering us — is what sustains a medical technology company over the long term.
• Confirmation of where the sector is heading. Radiology in Colombia no longer asks whether it will use artificial intelligence. It asks how to fit it into the workflow without breaking what works, without diluting clinical responsibility and without the cost running away. That is exactly the question we are working on.

Thank you to the seminar organizers, to everyone who came to the stand with hard questions, and to the institutions that trust us and let us show real results. See you at the next one.

Seminar gallery

---

Want to see what we showed at the seminar in your own institution? Message us on WhatsApp from any page on the site and we will arrange a personalized demonstration of our artificial intelligence layer and the full RIS/PACS.