Coronary disease & ACSDiagnostic or prognostic study
General Purpose Deep Learning Attenuation Correction Improves Diagnostic Accuracy of Single-photon Emission Computed Tomography (SPECT) Myocardial Perfusion Imaging (MPI)
- Diagnostic accuracy study
- multicentre
- external validation
- Deep learning (DL) model developed in 4,894 patients from 4 sites
Authors' conclusion
In a multicenter external cohort, DeepAC improved prediction performance for obstructive CAD.
Study context
- Population
- Patients undergoing SPECT MPI; external validation in a clinical trial cohort (72 sites)first external cohort
- Participants
- 746 patients
- Comparison
- DeepAC vs nonattenuation correction (NC) total perfusion deficit (TPD)
- Outcome
- Diagnostic accuracy of TPD for obstructive CAD (left main stenosis ≥50% or ≥70% in other vessels)
Performance
As reported in the abstract
AUC for obstructive CAD
| Test or model | AUC |
|---|---|
| DeepAC | 0.77 (95% CI: 0.73-0.81) |
| NC | 0.73 (95% CI: 0.69-0.77) |
- DeepAC vs NC: P < 0.001
Further findings
Agreement with actual attenuation correction (AC) scores (second external cohort, n = 320)
- DeepAC quantitative scores had closer agreement with actual AC scores than NC
Original publication
General Purpose Deep Learning Attenuation Correction Improves Diagnostic Accuracy of SPECT MPI. JACC Cardiovasc Imaging. Published 8 August 2025. doi:10.1016/j.jcmg.2025.06.010
Summarised from the published abstract; every number is checked against it