AI Learning for Pediatric Right Ventricular Assessment: Development and Validation Across Multiple Centers

Charitha D. Reddy; Yi Yan; Min Qiu; Yi Tang; Bo Jin; Zhi Han; Yuhang Li; Sihan Zhou; Qiming Tang; Huan Xiao; Yang Shu; Lanping Wu; Lijun Fu; Y. F. Yang; Yuqi Zhang; Takumi Ichikawa; Ronald J. Wong; Ivana Marić; Nima Aghaeepour; Brice Gaudillière; Martin S. Angst; Karl G. Sylvester; Harvey J. Cohen; Gary L. Darmstadt; David K. Stevenson; Henry Chubb; Scott R. Ceresnak; Animesh Tandon; Doff B. McElhinney; Hao Zhang; Xuefeng B. Ling

doi:10.1101/2025.03.14.25323989

Why the study?

Accurate RV assessment in children is challenging due to complex geometry and morphological variability, prompting the need for automated, standardized AI assessments to improve diagnosis and management.

Does an AI learning framework accurately automate right ventricular assessment and identify pulmonary hypertension in pediatric patients?

Key Points

To develop and validate a multi-task artificial intelligence framework for automated segmentation and functional assessment of the right ventricle in pediatric echocardiograms.
Trained and evaluated a multi-task deep learning model using 24,984 echocardiograms from 3,993 pediatric patients across four tertiary centers located in North America (one center) and Asia (three centers).
Model evaluated for ventricular segmentation, beat-by-beat quantification of right ventricular fractional area change (FAC), pulmonary hypertension classification, and left ventricular functional assessment.
The system achieved Dice similarity coefficients of 0.86 in apical-four-chamber and 0.88 in parasternal-short-axis views for right ventricular segmentation compared to expert manual annotations.
Right ventricular functional assessment achieved an AUC of 0.95 in the U.S. cohort and 0.97 in the Asian cohort.
Diagnostic accuracy for pulmonary hypertension classification was 0.95 in the U.S. cohort and 0.94 in the Asian cohort, with the framework also improving left ventricular ejection fraction predictions across both cohorts.

Structured PICO

Does an AI learning framework accurately automate right ventricular assessment and identify pulmonary hypertension in pediatric patients?

P

Population

3,993 pediatric patients (24,984 echocardiograms) across four tertiary care centers (one in North America, three in Asia). Median age 9.6 years. Diagnoses included normal, pulmonary hypertension (PH), tetralogy of Fallot (TOF), and others.

I

Intervention

AI framework using multi-task learning (U2-Net-based deep learning) for automated right ventricular (RV) segmentation, beat-by-beat quantification of RV fractional area change (FAC), identification of cardiac abnormalities like pulmonary hypertension (PH), and left ventricular (LV) functional assessment.

C

Comparator

Expert annotations and traditional manual echocardiographic assessment.

O

Outcome

Accuracy of RV segmentation (Dice similarity coefficients), RV functional assessment (AUC), and PH classification (diagnostic accuracy).surrogate

A novel AI framework accurately automates pediatric right ventricular segmentation, functional assessment, and pulmonary hypertension classification across diverse multinational cohorts.

Limitations

Algorithm trained on high-quality echocardiographic videos from a U.S. academic center, which may not represent real-world variability
Broader validation is needed, particularly in low-resource settings
LV assessments were secondary to RV analysis and warrant further investigation

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

May support more consistent pediatric RV assessment where expertise is limited; leaves open outcome trials before changing practice.

Abstract

Background: Congenital and acquired heart disease affects ∼1% of children globally, with right ventricular (RV) dysfunction being a common and complex issue due to conditions like congenital heart disease (CHD), pulmonary hypertension (PH), and prematurity. Accurate RV assessment is challenging due to its unique geometry, interventricular interactions, and morphological variability in pediatric patients. Fractional area change (FAC), a key echocardiographic measure, correlates strongly with disease severity, aiding in timely intervention and prognosis. AI learning shows the potential to automate and standardize RV assessments, overcoming traditional limitations and improving early diagnosis and management of pediatric cardiovascular disorders. Methods: Using 24,984 echocardiograms from 3,993 pediatric patients across four tertiary care centers (one in North America, three in Asia), we developed and validated an AI framework for automated RV assessment. The framework employs multi-task learning to perform ventricular segmentation, beat-by-beat quantification of RV FAC, and identification of cardiac abnormalities like PH. It was also extended to enhance left ventricular (LV) functional assessment. Findings: Our AI system achieved Dice similarity coefficients of 0.86 (apical-four-chamber, A4C) and 0.88 (parasternal-short-axis, PSAX) for RV segmentation, matching expert annotations. It demonstrated robust RV functional assessment, with AUCs of 0.95 (U.S. cohort) and 0.97 (Asian cohort). For PH classification, diagnostic accuracies were 0.95 (U.S.) and 0.94 (Asian), confirming consistent performance across populations. When extended to LV assessment, the framework significantly improved LV ejection fraction (EF) prediction in both U.S. and Asian cohorts. Interpretation: This validated AI framework enables reliable, automated ventricular function analysis, matching expert-level performance. By enhancing clinical workflows and standardizing pediatric cardiac assessments, it has the potential to improve care management for pediatric cardiovascular disorders, particularly in resource-limited settings. Funding: This work was supported by the U.S. NIH 1R41HL160362-01 to XBL and K23HL150279 to AT.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Reddy et al. (2025) studied this question.

synapsesocial.com/papers/6a7f7c2f53b9c9cf1aa5562b https://doi.org/10.1101/2025.03.14.25323989

View Full Paper

Ask AI

Bookmark

Share

AI Learning for Pediatric Right Ventricular Assessment: Development and Validation Across Multiple Centers

Key Points

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

AI Learning for Pediatric Right Ventricular Assessment: Development and Validation Across Multiple Centers

Key Points

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading