Follow-up of recommendations in imaging reports
Every recommendation, owned.
Designed to read each signed report, quote the clinician's own sentence and hand the recommendation to the department the hospital has configured as responsible, with a named person and a due date. Antevion never writes for the clinician and makes no clinical decisions.
Fig. 1 — One sentence from a report, as Antevion records it
CT pulmonary angiogram · Emergency · 12/03
«Nódulo de 6 mm no lobo superior direito; recomenda-se TC de controlo1 em 6 a 12 meses2.»
Translation: “6 mm nodule in the right upper lobe; a control CT is recommended in 6 to 12 months.”
Everything Antevion records comes from the words in the report. It never suggests an interval.
The problem
Hospitals don't lack detection. They lack anyone who holds the state “recommended but not yet done”.
The report goes to the doctor who ordered the scan, often for an unrelated reason. The recommendation is prose in a document, not a task in anyone's queue. Nobody can count what is outstanding.
The loss concentrates between departments: a nodule on an emergency CT pulmonary angiogram, or on an abdominal CT, that belongs to pulmonology.
Two paths for the same report. Today: the report with the recommendation goes to the emergency doctor, the patient is discharged and pulmonology never receives the recommendation. With Antevion: the sentence is recorded, it goes to pulmonology with a named person and a due date, a later exam is proposed and a member of staff confirms.
Day 0 · CT pulmonary angiogram ordered by the emergency department
«Nódulo de 6 mm no lobo superior direito; recomenda-se TC de controlo em 6 a 12 meses.»
Translation: “6 mm nodule in the right upper lobe; a control CT is recommended in 6 to 12 months.”
Today
Day 1
The report goes to the emergency doctor. The embolism is ruled out and the patient is discharged.
Week 1
Pulmonology never receives the recommendation.
Month 12
The due date arrives. No reminder, no list it appears on.
After
The scan is never done, and nobody can see that it is outstanding.
No follow-up
With Antevion
Day 1
The recommendation is handed to pulmonology, with a named person.
Week 1
The due date is on pulmonology’s list, taken from the clinician’s interval.
Month 12
A chest CT that was performed is proposed as a possible fulfilment.
After
A member of staff confirms. Nothing closes on its own.
Completed, confirmed by a person
Of 1,910 CT pulmonary angiograms at three Toronto teaching hospitals, 35.3% showed lung nodules. Only 25.1% of patients with new nodules needing follow-up got an explicit follow-up recommendation, and 53.8% of those were not re-imaged on time. 1
In a US safety-net cohort of 551 patients with incidental 5–8 mm lung nodules, only 28% completed guideline surveillance; 39% had none documented. 2
Portugal performs more CT scans per person than any other EU country: 29,723 per 100,000 inhabitants (hospitals only). 3
None of this is anyone's fault. It's structural.
How it works
Follow one fictitious case from report to confirmation. The clinician’s sentence travels intact; a wrong proposal is rejected; only a person closes it.
One case, from the report to confirmation
Illustrative example with fictitious data. It does not represent a real patient, clinician or hospital. Dates, department and sentence are fictional. Criteria are shown as “met” or “not met”, never as a score.
1 · Report sentence
“Nódulo pulmonar no lobo inferior direito. Sugere-se TC de tórax de controlo dentro de 6 meses.”
Translation: “Pulmonary nodule in the right lower lobe. A control chest CT within 6 months is suggested.”
Recorded: action “chest CT” · interval “6 months” — both quoted from the sentence
2 · Owning department
Pulmonology — configured by the hospital for this type of finding
Owner: pulmonology coordinator
Due: 4 Sep — from the interval the clinician stated
3 · Proposed fulfilments
Chest radiograph · 12 May
Same patient — met
Exam type named in the recommendation — not met
After the report — met
Rejected by the pulmonology coordinator: a different type of exam from the one recommended. Nothing closes. The reason is recorded.
Does not close
Chest CT · 21 Aug
Same patient — met
Exam type named in the recommendation — met
After the report — met
Awaiting confirmation
4 · Confirmed by a person
Confirmed by the pulmonology coordinator · reason: recommended exam performed
State: completed
Illustrative example. A sentence from a CT pulmonary angiogram report is recorded; the recommendation goes to pulmonology, due 4 September. A chest radiograph is proposed and rejected because it is not the recommended type of exam — nothing closes. A chest CT is proposed, confirmed by a person, and the follow-up is completed.
It reads each signed report of the scan types in scope and records the follow-up recommendation the clinician wrote, quoting the clinician's own sentence.
It hands the recommendation to the department the hospital has configured as responsible for that type of finding, with a named person and a due date taken from the interval the clinician stated.
When a recommendation lacks an interval or an action, it asks the report's author to complete it. It never fills it in.
It proposes the later exam or appointment that may fulfil the recommendation, for a member of staff to confirm. Nothing closes automatically. A near-miss is rejected and nothing closes.
It lets the patient tell the hospital that the follow-up was done at another provider. Staff verify it before anything changes.
It reports, by department pair, how many recommendations were completed, completed elsewhere, closed as not indicated, overdue, or never actioned, and always states its coverage.
These steps describe how Antevion is designed to work.
Limits
These limits are deliberate.
The audit
A retrospective audit of 12 months of reports, before any workflow changes.
Retrospective follow-up audit
Recommendations with a stated interval
(no data: filled by the audit)
No follow-up found, by department pair
Illustrative example of an audit report, without numbers: an empty bar with two categories and two rows by department pair.
Fee on request.
Request a leakage auditBy role
Antevion asks; it never writes for the radiologist. Incomplete recommendations go back to their author. Nothing is pre-filled or suggested.
Question to the author
“Nódulo no lobo superior esquerdo; sugere-se controlo imagiológico.” Translation: “Nodule in the left upper lobe; imaging follow-up suggested.”
Missing: exam type · interval
The fields start empty. There is no default.
Illustrative example of a question to the author: the report sentence, the missing fields (exam type and interval) left empty, and two options — record follow-up or no follow-up intended.
Follow-ups left undone concentrate between departments. The audit shows, for each department pair — who ordered the scan and who owns the finding — how many recommended follow-ups have a matching later exam or appointment, and how many have no follow-up found.
By department pair · pulmonology as owning department
| Ordering department | Matching later exam or appointment found (unverified) | No follow-up found |
|---|---|---|
| Emergency | (no data: filled by the audit) | (no data: filled by the audit) |
| Internal medicine | (no data: filled by the audit) | (no data: filled by the audit) |
| Cardiology | (no data: filled by the audit) | (no data: filled by the audit) |
| General surgery | (no data: filled by the audit) | (no data: filled by the audit) |
Coverage: scan types in scope · period · reports processed against expected
Your audit fills this table with your hospital's numbers.
Example table with no numbers: ordering departments (Emergency, Internal medicine, Cardiology, General surgery) against two audit categories — matching later exam or appointment found (unverified) and no follow-up found — with pulmonology as owning department. The audit fills the cells.
Antevion is designed so that every closure records who made it and why, and every report states its coverage — a verifiable record your quality and risk committee can review.
Closure record
Designed to be hosted in the EU, single-tenant, with the hospital as data controller under a GDPR Article 28 agreement. No real patient data before a signed data-processing agreement and a DPIA.
The data path in the audit
Before any real data: a signed data-processing agreement and a DPIA.
The data path in the audit: a file export from the hospital, a single-tenant environment designed for EU hosting, and a report delivered to the hospital; the hospital is the controller and Antevion the processor.
The audit is a fixed fee, sized for a direct award, with about 6 weeks of analysis once the data is available. It requires no change to anyone's workflow.
See the full auditThe terms at a glance
Evidence
A community-hospital lung-nodule tracking service saw 78% of tracked patients complete follow-up, versus 52% of patients never reported to it. 4
Across 31 studies of interventions to improve lung-nodule follow-up, tracking systems appeared the most effective — though most studies carried a serious risk of bias. 5
The audit measures this in your hospital: how many recommended follow-ups have a matching later exam or appointment in your own records.
Regulation and data
Antevion does not analyse images, estimate risk or propose treatment. Antevion never generates an interval and never suggests a recommendation.
Clinicians make every clinical decision. A person confirms every closure, with a reason.
We do not yet have a written qualification opinion on Antevion under the EU Medical Device Regulation. Until we do, we make no claim about its status. We will obtain one before any pilot with real patient data.
This is Antevion's intended-purpose text. Nothing we say in public goes beyond it.
Intended purpose. Antevion is administrative workflow software for hospitals. It records follow-up recommendations that clinicians have written in imaging reports, quoting the report text on which each record is based. It routes each recorded recommendation to the department that the hospital has configured as responsible for that type of finding, and assigns an owner.
It calculates due dates only from intervals stated by a clinician. Antevion never generates an interval. Where a recommendation states no interval, Antevion opens the follow-up recommendation immediately in the owning department as "no interval recorded" and, at the same time, asks the report author to state one. If the author has not answered within the number of days the hospital has configured, a named clinician in the owning department sets the interval. Where a recommendation lacks an action, Antevion asks the author to complete it.
Where the hospital enables it, Antevion also asks the report author to state whether follow-up is intended when a report mentions a finding type for which the hospital's reporting policy requires a follow-up statement and none was found.
Antevion identifies later scheduling events, namely a performed examination or an appointment of the kind the recommendation names, that may fulfil a recommendation, for staff to confirm. It sends messages to responsible staff and family doctors and, through the hospital's own channels and in wording approved by the hospital, to patients. It reports completion rates.
Antevion does not analyse images, assess the clinical significance of findings, generate or suggest clinical recommendations or intervals, estimate risk, prioritise patients on clinical criteria, monitor treatment, or propose treatment. All clinical decisions are made by clinicians.
Platform
Antevion is designed as an accountability layer for what clinicians write: it quotes the clinician's words, gives each intention an owner and proves whether it happened. Imaging follow-up is the product in development: it is where this core starts, not where it ends.
The clinician's exact words, never paraphrased and never generated.
A named owner and a due date taken from the interval the clinician stated.
A person confirms that it happened, with a reason. Nothing closes on its own.
What was never done, by department pair, with its coverage.
Further products are planned on the same core, each with its own regulatory review.
Products
Every product is designed on the same principles: it quotes the clinician, never generates an interval or a verdict, and a person decides.
Imaging follow-up
Designed to quote each follow-up recommendation in a signed imaging report and give it an owner and a due date. It starts with a leakage audit of the hospital's own reports.
Screening and suspected cancer
All signed reports
Trial pre-screening
Inherited cancer risk
Evidence and trials
Planned products are plans, not current operations. Each will be subject to its own regulatory review before any pilot.
See all productsQuestions
Yes. Hospitals have done it with a dedicated navigator and manual processes, such as Nodule Net. 4 These depend on someone reporting each patient and on staff finding each completion, and a single hospital cannot see follow-up done at another provider. The audit shows where your own process loses follow-ups.
At a hospital that combined tracking with other quality interventions, radiologist recommendations for extra imaging fell 44% (10% to 5.6% of reports) while staying flat at a control hospital, and recommendations that specified what to do rose from 5.6% to 42.3%. 6 Antevion never generates or suggests recommendations; it records only the ones clinicians write.
Antevion works from what clinicians write. Finding an unreported nodule would mean analysing images, which is outside Antevion's intended purpose. Antevion is designed as a vendor-neutral ledger for follow-ups a clinician recorded; we plan to take in the output of detection tools the hospital already uses, but this is not built.
Those tools decide or detect. Antevion carries what a clinician already wrote, across departments and providers. Taking in their output as one more input is planned, not built.
A named coordinator in the owning department, with protected time agreed in the pilot terms. The worklist is designed to be cleared in about 20 minutes a day.
By a person, always, with a reason. A near-miss is rejected and nothing closes — the diagram in “How it works” shows one.
We do not yet have a written qualification opinion on Antevion under the EU Medical Device Regulation. Until we do, we make no claim about its status. We will obtain one before any pilot with real patient data. Its intended purpose is set out in “Regulation and data”.
Contact
Write to us with your role, your institution and the departments involved. We reply by email and propose a conversation about the audit. Please never include patient data.
Every number on this page comes from a published source. We give each study's setting and year.
Kwan JL, Yermak D, Markell L, Paul NS, Shojania KG, Cram P. Follow Up of Incidental High-Risk Pulmonary Nodules on Computed Tomography Pulmonary Angiography at Care Transitions. J Hosp Med. 2019;14(6):349–352.
Three Toronto teaching hospitals (Canada), scans from 2014–2015; a single study. The need for follow-up was judged by the authors.
Lee JS, Lisker S, Vittinghoff E, et al. Follow-up of incidental pulmonary nodules and association with mortality in a safety-net cohort. Diagnosis (Berl). 2019;6(4):351–359.
San Francisco safety-net network (US), 2008–2016. In this study, gaps in surveillance were not significantly associated with mortality.
Eurostat. Healthcare resource statistics – technical resources and medical technology. Statistics Explained (data extracted January 2026).
Examinations in hospitals only; 2023 data.
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Healthcare_resource_statistics_-_technical_resources_and_medical_technology Back to text
Singh H, Koster M, Jani C, et al. Nodule net: A centralized prospective lung nodule tracking and safety-net program. Respir Med. 2022;192:106737.
One US community hospital (Cambridge, Massachusetts). Non-randomised comparison: patients not reported to the service may differ.
Aunger J, Yip KP, Dosanjh K, et al. Interventions to Improve Adherence to Clinical Guidelines for the Management and Follow-Up of Pulmonary Nodules: A Systematic Review. Chest. 2025;168(1):248–268.
Systematic review; pulmonary nodules only. The authors state their conclusion with caution.
Abbasi N, Kapoor N, Lacson R, et al. Cumulative Effect of Targeted Interventions on Radiologist Recommendations for Additional Imaging. Radiology. 2025;315(3):e243750.
Two Boston tertiary hospitals (US), non-randomised, over 8 years. The intervention combined education, tracking tools, added staff and performance reports.