Automated segmentation of visceral, deep subcutaneous, and superficial subcutaneous adipose tissue volumes in MRI of neonates and young children

Abstract

Purpose: to develop and evaluate an automated segmentation method for accurate quantification of abdominal adipose tissue (AAT) depots (superficial subcutaneous [SSAT], deep subcutaneous [DSAT], and visceral adipose tissue [VAT]) in neonates and young children.

Methods: this was a secondary analysis of prospectively collected data, which used abdominal MRI data from Growing Up in Singapore Towards healthy Outcomes (GUSTO), a longitudinal mother-offspring cohort to train and evaluate a convolutional neural network for volumetric AAT segmentation. The data comprised imaging volumes of 333 neonates obtained at early infancy (age ≤ 2 weeks, 180 males) and 755 children at ages 4.5 years (n = 316, 150 males), and 6 years (n = 439, 219 males). The network was trained on images from 761 randomly selected volumes (neonates and children combined) and evaluated on 100 neonatal volumes and 227 childhood volumes using 10-fold validation. Automated segmentations were compared to expert-generated manual segmentation. Segmentation performance was assessed using Dice scores.

Results: when the model was tested on the test datasets in the 10-folds, the model had strong agreement with the ground truth for all testing sets, with mean Dice similarity for SSAT, DSAT, and VAT, respectively, of 0.960, 0.909, and 0.872 in neonates and 0.944, 0.851, and 0.960 in children. The model generalized well to different body sizes and ages, as well as on all abdominal levels.

Conclusion: the proposed segmentation approach provided accurate automated volumetric assessment of AAT compartments on MRI from neonates and children.

More information

Identifiers

ORCID for Keith Godfrey: orcid.org/0000-0002-4643-0618

Catalogue record

Export record

Altmetrics