Deep learning for vessel segmentation and flow analysis to identify clusters associated with adverse outcomes in a fontan patient registry

We introduce a deep learning framework comprising two models for automated segmentation (DCS) and large-scale deep temporal clustering (DTC) within a registry of single ventricle patients. The DCS model performs simultaneous classification and segmentation of velocity-encoded phase-contrast magnetic resonance (PCMR) data for five individual blood vessels, the left and right pulmonary arteries, aorta, superior vena cava, and inferior vena cava. Trained, validated and tested on 260 cardiac MRI exams (each containing 5 PCMR scans), it demonstrated a median Dice score of 0.91 on 50 unseen test exams. Integrated into a fully automated pipeline, the DCS model processed over 4500 registry exams without manual intervention, reaching 98% classification accuracy and 90% segmentation accuracy in cases with all five vessels present. Flow curves obtained from successful segmentations were used to train the DTC model, which performs deep temporal clustering to uncover unique flow patterns. Survival analysis showed that these groups were statistically correlated to increased risk of mortality or transplantation and to liver disease, highlighting the clinical relevance of the proposed framework.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-03-04
DOI
https://doi.org/10.1038/s41598-026-40738-6
Primary Topic
Congenital Heart Disease Studies
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Deep learning for vessel segmentation and flow analysis to identify clusters associated with adverse outcomes in a fontan patient registry

Michael A. Quail, Pierangelo Renella, Aswathy Vaikom House, Christopher Z. Lam et al.
Scientific Reports
Congenital Heart Disease Studies
article

Deep learning for vessel segmentation and flow analysis to identify clusters associated with adverse outcomes in a fontan patient registry

Michael A. Quail, Pierangelo Renella, Aswathy Vaikom House, Christopher Z. Lam, Kenan W.D. Stern, Markus S. Renno, Timothy C. Slesnick, Jennifer Anne Steeden, Manish Aggarwal, Francesca Raimondi, Rajesh Krishnamurthy, Matthew D. Files, Jonathan H. Soslow, Adam L. Dorfman, Joshua Daniel Robinson, Andrew W. Hoyer, Ashish N. Doshi, Nicole E. St Clair, Gabriel F. Miller, Alison L. Marsden, Muthurangu Vivek, Tarek Alsaied, Amee M. Shah, Laura Olivieri, Preeti Ramachandran, Sanjeet Hegde, Garg Ruchira, Justin Weigand, Bijoy Thattaliyath, Mark A. Fogel, FORCE Investigators, Tiffanie Johnson, Tina Yao, Jeremy Steele, Madeline Gong, Yue-Hin Loke, Rahul H. Rathod
article en

Abstract

We introduce a deep learning framework comprising two models for automated segmentation (DCS) and large-scale deep temporal clustering (DTC) within a registry of single ventricle patients. The DCS model performs simultaneous classification and segmentation of velocity-encoded phase-contrast magnetic resonance (PCMR) data for five individual blood vessels, the left and right pulmonary arteries, aorta, superior vena cava, and inferior vena cava. Trained, validated and tested on 260 cardiac MRI exams (each containing 5 PCMR scans), it demonstrated a median Dice score of 0.91 on 50 unseen test exams. Integrated into a fully automated pipeline, the DCS model processed over 4500 registry exams without manual intervention, reaching 98% classification accuracy and 90% segmentation accuracy in cases with all five vessels present. Flow curves obtained from successful segmentations were used to train the DTC model, which performs deep temporal clustering to uncover unique flow patterns. Survival analysis showed that these groups were statistically correlated to increased risk of mortality or transplantation and to liver disease, highlighting the clinical relevance of the proposed framework.

Scientific Reports
Cedars-Sinai Medical Center (US), Boston Children's Hospital (US), Lurie Children's Hospital (US), Morgan Stanley Children's Hospital (US), Seattle Children's Hospital (US), Arkansas Children's Hospital (US), Children's Hospital of Orange County (US), Nationwide Children's Hospital (US), Yale New Haven Hospital (US), St. Louis Children's Hospital (US), Children's Hospital of Philadelphia (US), Children's National (US), University of Arizona (US), University of Kentucky (US), Emory University (US), Palo Alto University (US), Hospital for Sick Children (CA), Riley Hospital for Children (US), Rady Children's Hospital-San Diego (US), University of Iowa Stead Family Children’s Hospital (US), Ospedale Papa Giovanni XXIII (IT), Children's Hospital of Oklahoma (US), Children's Hospital of Pittsburgh (US), Texas Children's Hospital (US), Johns Hopkins Hospital (US), C. S. Mott Children's Hospital (US), University College London (GB), Vanderbilt University Medical Center (US), Stanford University (US)
Yale University, Vanderbilt University, Johns Hopkins University, Seattle Children's Research Institute, Cedars-Sinai Medical Center, Vanderbilt University Medical Center, Children's National Hospital
Good health and well-being
Openalex Percentile: Top 36%
Congenital Heart Disease Studies
DOIOpen Access PDFOpenAlex
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.