Artificial Intelligence - Based Opportunistic Coronary Artery Calcium Scoring on Routine Chest-CT Scan

Opportunistic Screening of Coronary Artery Calcium on Non-Gated Routine Chest CT Using Artificial Intelligence: Retrospective External Validation and Clinical Risk Stratification

This retrospective, non-interventional study externally validates a pre-trained open-weight deep-learning algorithm (Swin-UNETR) for the opportunistic quantification of coronary artery calcium (CAC) on non-gated routine chest CT scans acquired at a German academic center, and evaluates the prognostic value of this automated imaging biomarker for cardiovascular risk stratification. Coronary calcium is an established predictor of cardiovascular risk, but is not routinely quantified on the tens of thousands of non-cardiac chest CTs performed each year. Because existing high-performing AI models were trained almost exclusively on U.S. cohorts, external validation on a European scanner fleet is required to exclude scanner bias (domain shift). The study comprises three linked analytic cohorts: (1) a validation cohort comparing the AI-CAC score against the ECG-gated cardiac CT Agatston reference; (2) a dialysis cohort assessing calcification progression and mortality; and (3) an emergency department cohort assessing short-term cardiovascular events. This is an investigator-initiated trial with no intervention on patients.

Study Overview

• Status: Not yet recruiting
• Conditions: Coronary Artery Disease; Cardiovascular Risk; Coronary Artery Calcification; Emergency Care; End-Stage Renal Disease Requiring Haemodialysis

Detailed Description

The study analyzes a retrospective cohort of routine clinical CT examinations at University Hospital Cologne. Data originate from the hospital information system and Picture Archiving and Communication System (PACS) and are provided in pseudonymized form via the Medical Data Integration Center (MeDIC), acting as an independent trusted third party; the re-identification key remains under the sole control of MeDIC. Deep-learning inference is performed locally on isolated, access-controlled graphics processing unit (GPU) clusters of the institution (privacy by design / zero data retention); an open-weight model (Swin-UNETR) is used.

Three research questions are addressed in three analytic cohorts:

• Validation (n ≈ 150): diagnostic agreement of the automatically extracted AI-CAC score (from the non-gated CT) with the reference Agatston score from a paired ECG-gated cardiac CT acquired within ≤ 12 months.
• Dialysis (n ≈ 300): 150 hemodialysis patients plus 150 matched kidney-healthy controls with serial non-gated CTs; annualized calcification progression rate and all-cause mortality.
• Emergency department (n ≈ 1,500): patients > 50 years with non-gated chest CTs from the Emergency Department (without a primary cardiac focus); occurrence of in-hospital major adverse cardiac event (MACE) or cardiovascular readmission within 12 months.

Extracted data include demographics (age at examination, sex), cardiovascular risk factors and comorbidities (ICD-10), long-term medication, laboratory values, examination metadata (date, scanner manufacturer, kilovolt peak (kVp), slice thickness), and outcome data (mortality, cardiovascular events, readmissions). Statistical analysis uses Spearman correlation, Cohen's kappa and Bland-Altman analysis for method comparison; t-test / Mann-Whitney-U for group differences in progression; and Kaplan-Meier (log-rank) plus multivariable Cox proportional-hazards and logistic regression for outcome prediction. Legal basis: § 6 (1) no. 2 Health Data Use Act of Germany (GDNG) in conjunction with Art. 9 (2) (j) and Art. 89 (1) GDPR (research privilege); no individual consent (disproportionate effort, Art. 14 (5) (b) GDPR). The AI (artificial intelligence) model carries no CE-marking and is used strictly as a research tool; AI-CAC scores are not systematically fed back into clinical care.

• Study Type: Observational
• Enrollment (Estimated): 1950

Contacts and Locations

• Study Contact: Name: Cem Özel, MD; Phone Number: +49 176 2113 7580; Email: cem.oezel@uk-koeln.de
• Study Contact Backup: Name: Carsten Gietzen, MD; Email: carsten.gietzen@uk-koeln.de

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

Retrospective patients at University Hospital Cologne with routine clinical CT imaging (01 January 2015 - 31 December 2025), across three analytic cohorts (validation, dialysis, emergency department); approximately 1,950 cases.

Description

Inclusion Criteria:

• Validation cohort: patients with a paired non-gated chest CT and an ECG-gated cardiac CT acquired within ≤ 12 months.
• Dialysis cohort: hemodialysis patients with serial non-gated CTs, plus matched kidney-healthy controls.
• Emergency department cohort: patients > 50 years with non-gated chest CTs from the Emergency Department without a primary cardiac focus.

Exclusion Criteria:

• Documented objection to the scientific use of the data pursuant to Art. 21 GDPR.
• Cases lacking the minimum data required for analysis (insufficient image quality or missing reference/outcome data).

Study Plan

How is the study designed?

Number of groups / cohorts

3

Cohorts and Interventions

Group / Cohort
Validation Cohort; Paired non-gated chest CT and ECG-gated cardiac CT (≤ 12 months apart). AI-CAC score compared against the reference Agatston score. n ≈ 150.
Dialysis Cohort; 150 hemodialysis patients plus 150 matched kidney-healthy controls with serial non-gated CTs. Calcification progression and all-cause mortality. n ≈ 300.
Emergency Department Cohort; Patients > 50 years with non-gated chest CTs from the emergency department without primary cardiac focus. In-hospital MACE / cardiovascular readmission within 12 months. n ≈ 1,500

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Diagnostic agreement of the AI-CAC score with the reference Agatston score	Agreement between the automatically extracted AI-CAC score from the non-gated chest CT and the reference Agatston score from a paired ECG-gated cardiac CT, reported as Spearman correlation coefficient, Cohen's kappa across established risk classes (0, 1-100, 101-400, > 400), and Bland-Altman limits of agreement; supplemented by sensitivity, specificity, positive predictive value(PPV)/negative predictive value(NPV) and F1 score	At the index non-contrast chest CT (Day 0) and at the paired ECG-gated cardiac CT obtained within 12 months after the index CT.
Annualized calcification progression rate and all-cause mortality	Difference in the mean annual increase in AI-CAC between hemodialysis patients (dialysis cohort) and matched kidney-healthy controls measured on serial non-gated CTs (t-test / Mann-Whitney-U), and all-cause mortality analyzed by Kaplan-Meier (log-rank) and multivariable Cox proportional-hazards models (hazard ratios adjusted for confounders	From the index non-contrast chest CT (Day 0) through the last available serial non-gated chest CT and the end of individual follow-up, up to 10 years per participant.
In-hospital Major Adverse Cardiac Event or cardiovascular readmission within 12 months (Emergency Department cohort).	Occurrence of in-hospital Major Adverse Cardiovascular Events (myocardial infarction, stroke, resuscitation) or cardiovascular readmission within 12 months of the index Emergency Department visit, in relation to an unrecognized high AI calcium score (> 400); reported as adjusted odds ratios and hazard ratios from logistic regression and Cox models	12 months after the index emergency department visit

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Technical feasibility and inference time of the open-source AI model on local GPU clusters	Inference time per case and technical feasibility of running the open-weight deep-learning model (Swin-UNETR) as an isolated container on the institution's local GPU infrastructure	At the index non-contrast chest CT (Day 0)

Collaborators and Investigators

• Sponsor: University of Cologne
• Collaborators: Institute for Diagnostic and Interventional Radiology, University Hospital Cologne; Medical Data Integration Center (MeDIC), University Hospital Cologne
• Investigators: Principal Investigator: Cem Özel, MD, Department of Internal Medicine II, University Hospital Cologne

Study record dates

Study Major Dates

• Study Start (Estimated): November 1, 2026
• Primary Completion (Estimated): November 1, 2027
• Study Completion (Estimated): December 1, 2027

Study Registration Dates

• First Submitted: September 1, 2026
• First Submitted That Met QC Criteria: September 1, 2026
• First Posted (Actual): September 9, 2026

Study Record Updates

• Last Update Posted (Actual): September 9, 2026
• Last Update Submitted That Met QC Criteria: September 1, 2026
• Last Verified: September 1, 2026

More Information

Terms related to this study

• Keywords: Coronary Artery Disease; Artificial Intelligence; Hemodialysis; Deep Learning; Chest CT; Coronary Artery Calcification; MACE; Opportunistic Screening; Agatston Score; External Validation; Cardiovascular Risk Stratification
• Additional Relevant MeSH Terms: Vascular Diseases; Cardiovascular Diseases; Heart Diseases; Arteriosclerosis; Arterial Occlusive Diseases; Coronary Disease; Myocardial Ischemia; Coronary Artery Disease
• Other Study ID Numbers: AI-CAC-DE; 22-1182 (Other Identifier: Ethics Committee)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No