EBioMedicine. 2025;112:105560.

Artificial intelligence-enhanced comprehensive assessment of the aortic valve stenosis continuum in echocardiography.

, , , , , , , , , , ,

제1저자 박지석, 김지연 · 교신저자 전재익, 윤연이

PMID 39842286 PMC11794175 DOI 10.1016/j.ebiom.2025.105560 IF 11.2 순환기내과

소속 (7)

Cardiovascular Center and Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Gyeonggi, Republic of Korea; Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea.
CONNECT-AI Research Center, Yonsei University College of Medicine, Seoul, Republic of Korea; Department of Internal Medicine, Graduate School of Medical Science, Brain Korea 21 Project, Yonsei University College of Medicine, Seoul, Republic of Korea.
Ontact Health Inc., Seoul, Republic of Korea. Electronic address: jaeik.jeon@ontacthealth.com.
Cardiovascular Center and Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Gyeonggi, Republic of Korea; Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea; Ontact Health Inc., Seoul, Republic of Korea. Electronic address: yeonyeeyoon@snubh.org.
Ontact Health Inc., Seoul, Republic of Korea.
CONNECT-AI Research Center, Yonsei University College of Medicine, Seoul, Republic of Korea; Ontact Health Inc., Seoul, Republic of Korea.
CONNECT-AI Research Center, Yonsei University College of Medicine, Seoul, Republic of Korea; Ontact Health Inc., Seoul, Republic of Korea; Division of Cardiology, Severance Cardiovascular Hospital, Yonsei University College of Medicine, Yonsei University Health System, Seoul, Republic of Korea.

Abstract

BACKGROUND Transthoracic echocardiography (TTE) is the primary modality for diagnosing aortic stenosis (AS), yet it requires skilled operators and can be resource-intensive. We developed and validated an artificial intelligence (AI)-based system for evaluating AS that is effective in both resource-limited and advanced settings.

METHODS We created a dual-pathway AI system for AS evaluation using a nationwide echocardiographic dataset (developmental dataset, n = 8427): 1) a deep learning (DL)-based AS continuum assessment algorithm using limited 2D TTE videos, and 2) automating conventional AS evaluation. We performed internal (internal test dataset [ITDS], n = 841) and external validation (distinct hospital dataset [DHDS], n = 1696; temporally distinct dataset [TDDS], n = 772) for diagnostic value across various stages of AS and prognostic value for composite endpoints (cardiovascular death, heart failure, and aortic valve replacement).

FINDINGS The DL index for the AS continuum (DLi-ASc, range 0-100) increased with worsening AS severity and demonstrated excellent discrimination for any AS (AUC 0.91-0.99), significant AS (0.95-0.98), and severe AS (0.97-0.99). DLi-ASc was independent predictor for composite endpoint (adjusted hazard ratios 2.19, 1.64, and 1.61 per 10-point increase in ITDS, DHDS, and TDDS, respectively). Automatic measurement of conventional AS parameters demonstrated excellent correlation with manual measurement, resulting in high accuracy for AS staging (98.2% for ITDS, 82.1% for DHDS, and 96.8% for TDDS) and comparable prognostic value to manually-derived parameters.

INTERPRETATION The AI-based system provides accurate and prognostically valuable AS assessment, suitable for various clinical settings. Further validation studies are planned to confirm its effectiveness across diverse environments.

FUNDING This work was supported by a grant from the Institute of Information & Communications Technology Planning & Evaluation (IITP) funded by the Korea government (Ministry of Science and ICT; MSIT, Republic of Korea) (No. 2022000972, Development of a Flexible Mobile Healthcare Software Platform Using 5G MEC); and the Medical AI Clinic Program through the National IT Industry Promotion Agency (NIPA) funded by the MSIT, Republic of Korea (Grant No.: H0904-24-1002).

Keywords

Aortic stenosisArtificial intelligenceDiagnostic accuracyEchocardiographyPrognostic value

Figures

**Fig. 1** **AI-enhanced echocardiographic assessment of AS continuum.** The illustration depicts a dual-pathway AI system for evaluating AS. The top row illustrates the DL-based assessment of the AS continuum using limited views, providing a unique DL index for the AS continuum, termed DLi-ASc. The bottom row demonstrates the automated AS assessment, which derives conventional echocardiographic AS parameters. By integrating both pathways, our AI system enables accurate AS diagnosis and prognostication, making it broadly applicable in advanced and resource-limited settings. AS, aortic stenosis; DL, deep-learning; DLi-ASc, DL index for the AS continuum.

**Fig. 2** **The distribution of DLi-ASc according to AS severity and UMAP visualization.** (a) The DLi-ASc, generated by the DL-based AS continuum algorithm, showed a consistent trend of increasing scores with the progression of AS severity observed across both internal and external datasets. (b) The UMAP plot demonstrates a continuous nonlinear gradient transition from the normal state (grey) through AV sclerosis (yellow) to advanced AS stages (red), visually underscoring the DLi-ASc accurately representing the AS continuum. Abbreviations as in Fig. 1: DHDS, distinct hospital dataset; ITDS, internal test dataset; TDDS, temporally distinct dataset; UMAP, uniform manifold approximation and projection.

**Fig. 3** **Explainability analysis using saliency map.** The figure displays representative PLAX and PSAX views alongside their corresponding Grad-CAM saliency maps during DLi-ASc calculation. The saliency maps highlight that the DL model accurately focuses on the AV. As AS severity progresses from normal to severe AS, the model produces increasingly higher DLi-ASc scores, corresponding worsening AS. Abbreviations as in Fig. 1: AV, aortic valve; PLAX parasternal long-axis view; PSAX, parasternal short-axis view.

**Fig. 4** **Concordance in AS diagnosis between DL-based automated assessment and conventional evaluation.** (a) Across all datasets, the auto-measured AS parameters (AV maximal velocity, mean pressure gradient, and valve area) strongly correlated with those obtained from manual measurements. (b) Consequently, AS gradings from both methods exhibited a high concordance rate, ranging from 82.1% to 96.8%. Abbreviations as in Fig. 2, Fig. 3, Fig. 4: AVA, aortic valve area; LVOT, left ventricular outflow tract; MAE, mean absolute error; mPG, mean pressure gradient; r, Spearman Correlation Coefficient; V_max, maximal velocity.

**Fig. 5** **Diagnostic performances of DLi-ASc and other AI-derived conventional AS parameters across various stages.** The discriminative ability of DLi-ASc and other conventional AS parameters was consistently excellent for diagnosing any AS, significant AS (moderate to severe), and severe AS across all datasets: (a) ITDS, (b) DHDS, and (c) TDDS. Abbreviations as in Fig. 2, Fig. 4: AUC, the area under the curve; DLi-ASc, DL index for the AS continuum.

**Fig. 6** **Spline curves for composite outcomes associated with DLi-ASc.** The risk of composite outcome gradually increased with higher DLi-ASc across all datasets: (a) ITDS, (b) DHDS, and (c) TDDS. The solid lines represent the hazard ratio, and the blue shaded area represents the 95% confidence interval. Abbreviations as in Fig. 2, Fig. 4.

**Fig. 7** **Prognostic value of DLi-ASc and AS parameters.** The DLi-ASc showed independent predictive value for composite outcomes. Similarly, other AI-derived AS parameters were significant predictors for composite outcomes as well as manually-derived AS parameters. Abbreviations as in Fig. 2, Fig. 3, Fig. 4: HR, hazard ratio.

그림 출처: PubMed Central 공개 논문 (PMC11794175)

Conflict of interest statement

Declaration of interests Y.E.Y, J.J., Y.J., Y.H., and S.A.L. are currently affiliated with Ontact Health, Inc. J.J., J.K., and S.A.L are co-inventors on a patent related to this work filed by Ontact Health (Method For Providing Information On Severity Of Aortic Stenosis And Device Using The Same). H.J.C. holds stock in Ontact Health, Inc. The other authors have no conflicts of interest to declare.

그 밖의 정보