What does this research mean for the field?

A high-risk radiomic phenotype of epicardial adipose tissue, characterized by low volume and high spatial heterogeneity, independently predicts major adverse cardiovascular events. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This research aims to develop and validate an automated deep learning model for segmenting epicardial adipose tissue and estimating its impact on cardiovascular risk.

July 15, 2026Journal of Imaging Informatics in Medicine

A low-volume, highly heterogeneous EAT radiomic phenotype predicts ~82% greater MACE risk.

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Coronary artery disease risk (n=286). Intervention: High-risk radiomic phenotype of epicardial adipose tissue. Primary outcome: Major adverse cardiovascular events (MACE) (HR 1.82, 95% CI 1.04-3.17, p=0.035).

View Full Paper

Ask AI

Bookmark

Why the study?

Manual epicardial adipose tissue volume measurement on cardiac CT is slow, requires expertise, and has inter-reader variability, limiting its clinical utility for risk assessment.

Does a deep learning-derived epicardial adipose tissue radiomic phenotype from CCTA predict MACE?

Population

286 CCTA images of patients aged 40 to 90

Comparison

Automated 3D Residual U-Net segmentation vs manual measurements

Follow-up

2 years

Key result

A high-risk radiomic phenotype of epicardial adipose tissue, characterized by low volume and high spatial heterogeneity, independently predicted MACE (adjusted HR 1.82; 95% CI 1.04-3.17; p=0.035).

Authors

SNSepehr Nayebirad SFShayan Forghani SNSoroush Nematollahi

Discussion

Member takes

Overview

May refine MACE risk stratification on CCTA beyond fat volume alone; hypothesis-generating in this observational cohort.

Key Points

This research aims to develop and validate an automated deep learning model for segmenting epicardial adipose tissue and estimating its impact on cardiovascular risk.
286 coronary CT angiography images were analyzed, including training, validation, and testing sets with 60, 20, and 20 images respectively.
A 3D Residual U-Net model was trained using PyTorch and MONAI, optimized with Dice loss.
The association of model-derived epicardial fat volume with major adverse cardiac events was assessed using multivariable Cox proportional hazards regression.
The model segmented epicardial adipose tissue with a Dice score of 0.85 on unseen test scans.
Automated and manual epicardial fat volume measurements showed strong agreement (Spearman r = 0.932, p < 0.0001; CCC = 0.943, 95% CI 0.864–0.977).
While epicardial fat volume alone did not significantly predict major adverse cardiac events, a high-risk radiomic phenotype was identified as a strong independent predictor (adjusted HR 1.82, 95% CI 1.04–3.17, p = 0.035).

Study Design

Type

Cohort (n=286)

Structured PICO

Does a deep learning-derived epicardial adipose tissue radiomic phenotype from CCTA predict MACE?

Population

286 patients aged 40 to 90 undergoing coronary computed tomography angiography, followed for 2 years to assess major adverse cardiovascular events.

Exposure

Automated deep learning segmentation (3D Residual U-Net) of epicardial adipose tissue (EAT) to estimate epicardial fat volume (EFV) and radiomic phenotype

Comparator

Manual expert measurements

Outcome

Segmentation accuracy (Dice similarity coefficient, Lin's concordance correlation coefficient, Spearman correlation) and association of model-derived metrics with major adverse cardiovascular events (MACE) over 2 years of follow-upsurrogate

Main Result

Hazard Ratio: 1.82 (95% CI 1.04–3.17)

p-value: p=0.035

Automated deep learning can accurately segment epicardial adipose tissue from CCTA, revealing that a specific radiomic phenotype (low volume, high heterogeneity) independently predicts MACE, whereas sheer volume does not.

Cite This Study

Nayebirad et al. (2026) conducted a cohort in Coronary artery disease risk (n=286). High-risk radiomic phenotype of epicardial adipose tissue was evaluated on Major adverse cardiovascular events (MACE) (HR 1.82, 95% CI 1.04-3.17, p=0.035). A high-risk radiomic phenotype of epicardial adipose tissue, characterized by low volume and high spatial heterogeneity, independently predicted MACE (adjusted HR 1.82; 95% CI 1.04-3.17; p=0.035).

synapsesocial.com/papers/6a57250788b21df875480de9 https://doi.org/10.1007/s10278-026-02106-8

View Full Paper

Ask AI

Bookmark

July 15, 2026Journal of Imaging Informatics in Medicine

A low-volume, highly heterogeneous EAT radiomic phenotype predicts ~82% greater MACE risk.

View Full Paper

Ask AI

Bookmark

Why the study?

Manual epicardial adipose tissue volume measurement on cardiac CT is slow, requires expertise, and has inter-reader variability, limiting its clinical utility for risk assessment.

Does a deep learning-derived epicardial adipose tissue radiomic phenotype from CCTA predict MACE?

Population

286 CCTA images of patients aged 40 to 90

Comparison

Automated 3D Residual U-Net segmentation vs manual measurements

Follow-up

2 years

Key result

A high-risk radiomic phenotype of epicardial adipose tissue, characterized by low volume and high spatial heterogeneity, independently predicted MACE (adjusted HR 1.82; 95% CI 1.04-3.17; p=0.035).

Authors

SNSepehr Nayebirad SFShayan Forghani SNSoroush Nematollahi

Discussion

Member takes

Overview

May refine MACE risk stratification on CCTA beyond fat volume alone; hypothesis-generating in this observational cohort.

Key Points

This research aims to develop and validate an automated deep learning model for segmenting epicardial adipose tissue and estimating its impact on cardiovascular risk.
286 coronary CT angiography images were analyzed, including training, validation, and testing sets with 60, 20, and 20 images respectively.
A 3D Residual U-Net model was trained using PyTorch and MONAI, optimized with Dice loss.
The association of model-derived epicardial fat volume with major adverse cardiac events was assessed using multivariable Cox proportional hazards regression.
The model segmented epicardial adipose tissue with a Dice score of 0.85 on unseen test scans.
Automated and manual epicardial fat volume measurements showed strong agreement (Spearman r = 0.932, p < 0.0001; CCC = 0.943, 95% CI 0.864–0.977).
While epicardial fat volume alone did not significantly predict major adverse cardiac events, a high-risk radiomic phenotype was identified as a strong independent predictor (adjusted HR 1.82, 95% CI 1.04–3.17, p = 0.035).

Study Design

Type

Cohort (n=286)

Structured PICO

Does a deep learning-derived epicardial adipose tissue radiomic phenotype from CCTA predict MACE?

Population

286 patients aged 40 to 90 undergoing coronary computed tomography angiography, followed for 2 years to assess major adverse cardiovascular events.

Exposure

Automated deep learning segmentation (3D Residual U-Net) of epicardial adipose tissue (EAT) to estimate epicardial fat volume (EFV) and radiomic phenotype

Comparator

Manual expert measurements

Outcome

Main Result

Hazard Ratio: 1.82 (95% CI 1.04–3.17)

p-value: p=0.035

Cite This Study

synapsesocial.com/papers/6a57250788b21df875480de9 https://doi.org/10.1007/s10278-026-02106-8

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark