Abstract 4144057: AI-enabled Cardiac Chambers Volumetry in Non-Contrast Cardiac CT scans (AI-CAC) Detects HFrEF vs. HFpEF

Introduction: Coronary artery calcium (CAC) scans contain more information than is currently reported. We have previously shown in the Multi-Ethnic Study of Atherosclerosis (MESA) that AI-enabled left atrial (LA) volumetry in CAC scans (AI-CAC) enabled prediction of atrial fibrillation (AF) as early as one year. Furthermore, we have shown adding AI-CAC LA volumetry to CHA 2 DS 2 -VASc risk score improved stroke prediction in MESA. We have recently reported that AI-CAC left ventricular (LV) volumetry and mass significantly predicted incident heart failure (HF) and outperformed NT-proBNP. In this report, we examined whether AI-CAC can distinguish heart failure with reduced ejection fraction (HFrEF) versus heart failure with preserved ejection fraction (HFpEF). Methods: We applied the AutoChamber TM (HeartLung.AI, Houston, TX) component AI-CAC to data from 75 patients who underwent both a cardiac CT scan and echocardiography at Harbor UCLA medical center. AI-CAC took on average 21 seconds per scan and reported estimated volume for left atrium (LA), left ventricle (LV), right atrium (RA), and right ventricle (RV). LV volume was indexed (LVVI) by dividing LV volume by body surface area (BSA) to allow a more accurate assessment of the size of the left ventricle relative to the size of the individual. The average BSA for males and female were 1.59 ± 0.3 and 1.29 ± 0.2, respectively. HFrEF and HFpEF were defined as EF<40% and EF>50% respectively. Results: Average EF was 57.5 ± 7.0 in males and 59.7 ± 8.1 in females, respectively. AutoChamber volume for HFpEF vs HFrEF were LA (84.8+35.3 vs 113.2+32.8 p=0.002), LV (109.9+36.7 vs 170.7+65.9 p=0.0007), RA (97.5+58.3 vs 117.2+51.1 p=0.18), RV (135.6+52.1 vs 176.2+70.8 p=0.008), LVW (116.1+39.1 vs 170.6+56.9 p=0.0005). Density plots in the figure show a clear distinction between HFpEF and HFrEF using AI measured LVVI and comparable results. Conclusion: AI-enabled automated cardiac chambers volumetry in CAC scans correlates well with echocardiography-based LVVI and detects HFrEF vs. HFpEF. Further studies are needed to evaluate the clinical utility of AI-CAC and the ability of AutoChamber for prospective detection of patients at risk of HFrEF vs. HFpEF.

Authors

Institutions

Publication Details

Journal
Circulation
Published
2024-11-12
DOI
https://doi.org/10.1161/circ.150.suppl_1.4144057
Citations
1
Primary Topic
Advanced X-ray and CT Imaging
Type
article
Field-Weighted Citation Impact
0.18
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Abstract 4144057: AI-enabled Cardiac Chambers Volumetry in Non-Contrast Cardiac CT scans (AI-CAC) Detects HFrEF vs. HFpEF

Claudia I. Henschke, Emily Jaalouk, David F. Yankelevitz, Anthony P. Reeves et al.
1 citations
Circulation
Advanced X-ray and CT Imaging
0.18
article

Abstract 4144057: AI-enabled Cardiac Chambers Volumetry in Non-Contrast Cardiac CT scans (AI-CAC) Detects HFrEF vs. HFpEF

Claudia I. Henschke, Emily Jaalouk, David F. Yankelevitz, Anthony P. Reeves, Matthew Jay Budoff, K. Atlas, Chenyu Zhang, Morteza Naghavi, Sion Roy
article en
1 citations

Abstract

Introduction: Coronary artery calcium (CAC) scans contain more information than is currently reported. We have previously shown in the Multi-Ethnic Study of Atherosclerosis (MESA) that AI-enabled left atrial (LA) volumetry in CAC scans (AI-CAC) enabled prediction of atrial fibrillation (AF) as early as one year. Furthermore, we have shown adding AI-CAC LA volumetry to CHA 2 DS 2 -VASc risk score improved stroke prediction in MESA. We have recently reported that AI-CAC left ventricular (LV) volumetry and mass significantly predicted incident heart failure (HF) and outperformed NT-proBNP. In this report, we examined whether AI-CAC can distinguish heart failure with reduced ejection fraction (HFrEF) versus heart failure with preserved ejection fraction (HFpEF). Methods: We applied the AutoChamber TM (HeartLung.AI, Houston, TX) component AI-CAC to data from 75 patients who underwent both a cardiac CT scan and echocardiography at Harbor UCLA medical center. AI-CAC took on average 21 seconds per scan and reported estimated volume for left atrium (LA), left ventricle (LV), right atrium (RA), and right ventricle (RV). LV volume was indexed (LVVI) by dividing LV volume by body surface area (BSA) to allow a more accurate assessment of the size of the left ventricle relative to the size of the individual. The average BSA for males and female were 1.59 ± 0.3 and 1.29 ± 0.2, respectively. HFrEF and HFpEF were defined as EF<40% and EF>50% respectively. Results: Average EF was 57.5 ± 7.0 in males and 59.7 ± 8.1 in females, respectively. AutoChamber volume for HFpEF vs HFrEF were LA (84.8+35.3 vs 113.2+32.8 p=0.002), LV (109.9+36.7 vs 170.7+65.9 p=0.0007), RA (97.5+58.3 vs 117.2+51.1 p=0.18), RV (135.6+52.1 vs 176.2+70.8 p=0.008), LVW (116.1+39.1 vs 170.6+56.9 p=0.0005). Density plots in the figure show a clear distinction between HFpEF and HFrEF using AI measured LVVI and comparable results. Conclusion: AI-enabled automated cardiac chambers volumetry in CAC scans correlates well with echocardiography-based LVVI and detects HFrEF vs. HFpEF. Further studies are needed to evaluate the clinical utility of AI-CAC and the ability of AutoChamber for prospective detection of patients at risk of HFrEF vs. HFpEF.

CirculationVol. 150(Suppl_1)
HRL Laboratories (United States) (US), Cornell University (US), Mount Sinai Hospital (US)
Good health and well-being
Openalex Percentile: Top 51%
Advanced X-ray and CT Imaging
0.18
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.