PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
June 8, 2019European Heart Journal - Cardiovascular ImagingOpen Access

Diagnostic accuracy of 3D deep-learning-based fully automated estimation of patient-level minimum fractional flow reserve from coronary computed tomography angiography

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Deep-learning algorithms have been used to compute FFR from CCTA, but fully automated FFR calculation free from human input had not been achieved.

Does a fully automated 3D deep-learning model accurately estimate minimum FFR from CCTA data compared to invasive FFR in patients with coronary stenosis?

Population

1052 patients undergoing CCTA, including 131 with 30-90% stenosis undergoing invasive FFR

Comparison

Fully automated 3D deep-learning FFR vs visually determined CCTA >50% stenosis against invasive FFR

Design

Retrospective diagnostic accuracy study with Monte Carlo cross-validation

Authors

KKKanako K. KumamaruSFShinichiro FujimotoYOYujiro Otsuka

Discussion

Loading...

Member takes

Overview

May aid noninvasive FFR estimation from CCTA over visual stenosis grading; leaves open prospective validation before clinical adoption.

Structured PICO

Does a fully automated 3D deep-learning model accurately estimate minimum FFR from CCTA data compared to invasive FFR in patients with coronary stenosis?

P
Population
1052 patients, including 131 patients whose CCTA studies showed 30-90% stenosis and underwent invasive FFR (abnormal FFR observed in 72/131, 55%), and 921 patients who underwent clinically indicated CCTA without invasive FFR.
I
Intervention
Fully automated 3D deep-learning model for estimating minimum fractional flow reserve (FFR) from coronary computed tomography angiography (CCTA) data.
C
Comparator
Visually determined CCTA >50% stenosis, with invasive FFR as the reference standard.
O
Outcome
Diagnostic accuracy (AUC, sensitivity, specificity) for detection of abnormal FFR.surrogate

A fully automated 3D deep-learning model can estimate minimum FFR from CCTA data with moderate accuracy, significantly outperforming visual assessment of stenosis severity.

Cite This Study

Kumamaru et al. (2019) studied this question.

synapsesocial.com/papers/6a1e626928971e550d408fc1https://doi.org/10.1093/ehjci/jez160
View Full Paper
Ask AI
Bookmark
Share