Synapse
⌘+K
Synapse
PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
October 23, 2024Scientific ReportsOpen Access

An enhanced deep learning method for the quantification of epicardial adipose tissue

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Manual quantification of epicardial adipose tissue volume is labor-intensive and error-prone, while existing deep learning methods are mostly uninterpretable and fail to harness complete anatomical characteristics.

Does an enhanced deep learning method accurately quantify epicardial adipose tissue on CCTA compared to expert manual quantification?

Population

108 patients who underwent routine CCTA examinations

Comparison

Enhanced deep learning method vs expert manual quantification

Key result

An enhanced deep learning method for EAT quantification showed strong agreement with expert manual quantification, yielding a median 3D volume Dice score of 0.896 and correlation of 0.980 (p<0.001).

Authors

KTKe-Xin TangXLXiao‐Bo LiaoLYLing‐Qing Yuan

Discussion

Loading...

Member takes

Overview

Supports automated EAT quantification in research; leaves open clinical adoption pending prospective validation.

Structured PICO

Does an enhanced deep learning method accurately quantify epicardial adipose tissue on CCTA compared to expert manual quantification?

P
Population
108 patients who underwent routine CCTA examinations.
I
Intervention
Enhanced deep learning method integrating data-driven method and specific morphological information for epicardial adipose tissue (EAT) quantification
C
Comparator
Expert manual quantification
O
Outcome
Agreement between automatic method and expert manual quantification (Dice score coefficients, correlation, Bland-Altman bias)surrogate

Main Result

Effect estimate: correlation 0.980

p-value: p=<0.001

An enhanced deep learning method incorporating pericardial anatomical structures provides accurate automatic quantification of epicardial adipose tissue on CCTA scans.

Cite This Study

Tang et al. (2024) studied Patients undergoing routine CCTA examinations (n=108). Enhanced deep learning method vs. Expert manual quantification was evaluated on Agreement with expert manual quantification (3D volume Dice score coefficient and correlation) (correlation 0.980, p=<0.001). An enhanced deep learning method for EAT quantification showed strong agreement with expert manual quantification, yielding a median 3D volume Dice score of 0.896 and correlation of 0.980 (p<0.001).

synapsesocial.com/papers/6a956ec5ce3f9e00f3e3e065https://doi.org/10.1038/s41598-024-75659-9
View Full Paper
Ask AI
Bookmark
Share