PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
September 18, 2023BMC BioinformaticsOpen Access

Automatic segmentation of large-scale CT image datasets for detailed body composition analysis

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Manual segmentation of CT images for body composition analysis is time-consuming and subjective.

Population

More than 4000 CT subjects from the SCAPIS and IGT cohorts

Comparison

ResUNET vs UNET++ vs Ghost-UNET vs Ghost-UNET++

Design

Model development and validation study

Authors

NANouman AhmadRSRobin StrandBSBjörn Sparresäter

Discussion

Loading...

Member takes

Overview

Automated CT segmentation may speed body composition assessment; leaves open prospective validation for cardiometabolic risk stratification.

Structured PICO

P
Population
>4000 CT subjects from the SCAPIS and IGT cohort
I
Intervention
Fully automated segmentation techniques using convolutional neural network architectures (ResUNET, UNET++, Ghost-UNET, Ghost-UNET++) applied to a 3-slice CT imaging protocol
O
Outcome
Dice scores for automated segmentation of liver, spleen, skeletal muscle, bone marrow, cortical bone, and various adipose tissue depotssurrogate

Fully automated CNN-based segmentation, particularly UNET++, can efficiently and accurately analyze body composition from 3-slice CT scans in large-scale cohorts.

Cite This Study

Ahmad et al. (2023) studied this question.

synapsesocial.com/papers/6a7d1673b97b7d2d4f83d6a1https://doi.org/10.1186/s12859-023-05462-2
View Full Paper
Ask AI
Bookmark
Share