Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study (2024)

Wu, T., Estrada, S., van Gils, R., Su, R., Jaddoe, V. W. V., Oei, E. H. G., & Klein, S. (2024). Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study. American journal of Roentgenology, 222(1). https://doi.org/10.2214/AJR.23.29570

Wu, Tong ; Estrada, Santiago ; van Gils, Renza et al. / Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents : A Prospective Population-Based Study. In: American journal of Roentgenology. 2024 ; Vol. 222, Nr. 1.

@article{7df4a57fe0da4e74a2709a6cacaf0be4,

title = "Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study",

abstract = " BACKGROUND. The prevalence of childhood obesity has increased significantly worldwide, highlighting a need for accurate noninvasive quantification of body fat distribution in children. OBJECTIVE. The purpose of this study was to develop and test an automated deep learning method for subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) segmentation using Dixon MRI acquisitions in adolescents. METHODS. This study was embedded within the Generation R Study, a prospective population-based cohort study in Rotterdam, The Netherlands. The current study included 2989 children (1432 boys, 1557 girls; mean age, 13.5 years) who underwent investigational whole-body Dixon MRI after reaching the age of 13 years during the follow-up phase of the Generation R Study. A 2D competitive dense fully convolutional neural network model (2D-CDFNet) was trained from scratch to segment abdominal SAT and VAT using Dixon MRI-based images. The model underwent training, validation, and testing in 62, eight, and 15 children, respectively, who were selected by stratified random sampling, with manual segmentations used as reference. Segmentation performance was assessed using the Dice similarity coefficient and volumetric similarity. Two observers independently performed subjective visual assessments of automated segmentations in 504 children, selected by stratified random sampling, with undersegmentation and oversegmentation scored on a scale of 0-3 (with a score of 3 denoting nearly perfect segmentation). For 2820 children for whom complete data were available, Spearman correlation coefficients were computed among MRI measurements and BMI and dual-energy x-ray absorptiometry (DEXA)-based measurements. The model used (gitlab.com/radiology/msk/genr/abdomen/cdfnet) is publicly available. RESULTS. In the test dataset, the mean Dice similarity coefficient and mean volu-metric similarity, respectively, were 0.94 ± 0.03 [SD] and 0.98 ± 0.01 [SD] for SAT and 0.85 ± 0.05 and 0.92 ± 0.04 for VAT. The two observers assigned a score of 3 for SAT in 94% and 93% for the undersegmentation proportion and in 99% and 99% for the oversegmentation proportion, and they assigned a score of 3 for VAT in 99% and 99% for the undersegmentation proportion and in 95% and 97% for the oversegmentation proportion. Correlations with SAT and VAT were 0.808 and 0.698 for BMI and 0.941 and 0.801 for DEXA-derived fat mass. CONCLUSION. We trained and evaluated the 2D-CDFNet model on Dixon MRI in adolescents. Quantitative and qualitative measures of automated SAT and VAT segmentations indicated strong model performance. CLINICAL IMPACT. The automated model may facilitate large-scale studies investigating abdominal fat distribution on MRI among adolescents as well as associations of fat distribution with clinical outcomes. ",

keywords = "Male, Female, Humans, Child, Adolescent, Cohort Studies, Deep Learning, Prospective Studies, Pediatric Obesity, Abdominal Fat, Intra-Abdominal Fat, Magnetic Resonance Imaging/methods, Adipose Tissue, adolescents, MRI, deep learning, subcutaneous adipose tissue, visceral adipose tissue",

author = "Tong Wu and Santiago Estrada and {van Gils}, Renza and Ruisheng Su and Jaddoe, {Vincent W V} and Oei, {Edwin H.G.} and Stefan Klein",

year = "2024",

month = jan,

doi = "10.2214/AJR.23.29570",

language = "English",

volume = "222",

journal = "American journal of Roentgenology",

issn = "0361-803X",

publisher = "American Roentgen Ray Society",

number = "1",

}

Wu, T, Estrada, S, van Gils, R, Su, R, Jaddoe, VWV, Oei, EHG & Klein, S 2024, 'Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study', American journal of Roentgenology, vol. 222, nr. 1. https://doi.org/10.2214/AJR.23.29570

Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study. / Wu, Tong; Estrada, Santiago; van Gils, Renza et al.
In: American journal of Roentgenology, Vol. 222, Nr. 1, 01.2024.

Onderzoeksoutput: Bijdrage aan tijdschrift › Tijdschriftartikel › Academic › peer review

TY - JOUR

T1 - Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents

T2 - A Prospective Population-Based Study

AU - Wu, Tong

AU - Estrada, Santiago

AU - van Gils, Renza

AU - Su, Ruisheng

AU - Jaddoe, Vincent W V

AU - Oei, Edwin H.G.

AU - Klein, Stefan

PY - 2024/1

Y1 - 2024/1

N2 - BACKGROUND. The prevalence of childhood obesity has increased significantly worldwide, highlighting a need for accurate noninvasive quantification of body fat distribution in children. OBJECTIVE. The purpose of this study was to develop and test an automated deep learning method for subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) segmentation using Dixon MRI acquisitions in adolescents. METHODS. This study was embedded within the Generation R Study, a prospective population-based cohort study in Rotterdam, The Netherlands. The current study included 2989 children (1432 boys, 1557 girls; mean age, 13.5 years) who underwent investigational whole-body Dixon MRI after reaching the age of 13 years during the follow-up phase of the Generation R Study. A 2D competitive dense fully convolutional neural network model (2D-CDFNet) was trained from scratch to segment abdominal SAT and VAT using Dixon MRI-based images. The model underwent training, validation, and testing in 62, eight, and 15 children, respectively, who were selected by stratified random sampling, with manual segmentations used as reference. Segmentation performance was assessed using the Dice similarity coefficient and volumetric similarity. Two observers independently performed subjective visual assessments of automated segmentations in 504 children, selected by stratified random sampling, with undersegmentation and oversegmentation scored on a scale of 0-3 (with a score of 3 denoting nearly perfect segmentation). For 2820 children for whom complete data were available, Spearman correlation coefficients were computed among MRI measurements and BMI and dual-energy x-ray absorptiometry (DEXA)-based measurements. The model used (gitlab.com/radiology/msk/genr/abdomen/cdfnet) is publicly available. RESULTS. In the test dataset, the mean Dice similarity coefficient and mean volu-metric similarity, respectively, were 0.94 ± 0.03 [SD] and 0.98 ± 0.01 [SD] for SAT and 0.85 ± 0.05 and 0.92 ± 0.04 for VAT. The two observers assigned a score of 3 for SAT in 94% and 93% for the undersegmentation proportion and in 99% and 99% for the oversegmentation proportion, and they assigned a score of 3 for VAT in 99% and 99% for the undersegmentation proportion and in 95% and 97% for the oversegmentation proportion. Correlations with SAT and VAT were 0.808 and 0.698 for BMI and 0.941 and 0.801 for DEXA-derived fat mass. CONCLUSION. We trained and evaluated the 2D-CDFNet model on Dixon MRI in adolescents. Quantitative and qualitative measures of automated SAT and VAT segmentations indicated strong model performance. CLINICAL IMPACT. The automated model may facilitate large-scale studies investigating abdominal fat distribution on MRI among adolescents as well as associations of fat distribution with clinical outcomes.

AB - BACKGROUND. The prevalence of childhood obesity has increased significantly worldwide, highlighting a need for accurate noninvasive quantification of body fat distribution in children. OBJECTIVE. The purpose of this study was to develop and test an automated deep learning method for subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) segmentation using Dixon MRI acquisitions in adolescents. METHODS. This study was embedded within the Generation R Study, a prospective population-based cohort study in Rotterdam, The Netherlands. The current study included 2989 children (1432 boys, 1557 girls; mean age, 13.5 years) who underwent investigational whole-body Dixon MRI after reaching the age of 13 years during the follow-up phase of the Generation R Study. A 2D competitive dense fully convolutional neural network model (2D-CDFNet) was trained from scratch to segment abdominal SAT and VAT using Dixon MRI-based images. The model underwent training, validation, and testing in 62, eight, and 15 children, respectively, who were selected by stratified random sampling, with manual segmentations used as reference. Segmentation performance was assessed using the Dice similarity coefficient and volumetric similarity. Two observers independently performed subjective visual assessments of automated segmentations in 504 children, selected by stratified random sampling, with undersegmentation and oversegmentation scored on a scale of 0-3 (with a score of 3 denoting nearly perfect segmentation). For 2820 children for whom complete data were available, Spearman correlation coefficients were computed among MRI measurements and BMI and dual-energy x-ray absorptiometry (DEXA)-based measurements. The model used (gitlab.com/radiology/msk/genr/abdomen/cdfnet) is publicly available. RESULTS. In the test dataset, the mean Dice similarity coefficient and mean volu-metric similarity, respectively, were 0.94 ± 0.03 [SD] and 0.98 ± 0.01 [SD] for SAT and 0.85 ± 0.05 and 0.92 ± 0.04 for VAT. The two observers assigned a score of 3 for SAT in 94% and 93% for the undersegmentation proportion and in 99% and 99% for the oversegmentation proportion, and they assigned a score of 3 for VAT in 99% and 99% for the undersegmentation proportion and in 95% and 97% for the oversegmentation proportion. Correlations with SAT and VAT were 0.808 and 0.698 for BMI and 0.941 and 0.801 for DEXA-derived fat mass. CONCLUSION. We trained and evaluated the 2D-CDFNet model on Dixon MRI in adolescents. Quantitative and qualitative measures of automated SAT and VAT segmentations indicated strong model performance. CLINICAL IMPACT. The automated model may facilitate large-scale studies investigating abdominal fat distribution on MRI among adolescents as well as associations of fat distribution with clinical outcomes.

KW - Male

KW - Female

KW - Humans

KW - Child

KW - Adolescent

KW - Cohort Studies

KW - Deep Learning

KW - Prospective Studies

KW - Pediatric Obesity

KW - Abdominal Fat

KW - Intra-Abdominal Fat

KW - Magnetic Resonance Imaging/methods

KW - Adipose Tissue

KW - adolescents

KW - MRI

KW - deep learning

KW - subcutaneous adipose tissue

KW - visceral adipose tissue

UR - http://www.scopus.com/inward/record.url?scp=85184516323&partnerID=8YFLogxK

U2 - 10.2214/AJR.23.29570

DO - 10.2214/AJR.23.29570

M3 - Article

C2 - 37584508

SN - 0361-803X

VL - 222

JO - American journal of Roentgenology

JF - American journal of Roentgenology

IS - 1

ER -

Wu T, Estrada S, van Gils R, Su R, Jaddoe VWV, Oei EHG et al. Automated Deep Learning-Based Segmentation of Abdominal Adipose Tissue on Dixon MRI in Adolescents: A Prospective Population-Based Study. American journal of Roentgenology. 2024 jan.;222(1). doi: 10.2214/AJR.23.29570