Automated MRI quantification of pediatric abdominal adipose tissue using convolutional neural networks and novel total intensity maps

• Published in: Biomedical signal processing and control 2025-04, Vol.102, p.107250, Article 107250
• Main Authors: Suárez-García, José Gerardo, Alonso, Benito de Celis, Hernández-López, Javier Miguel, Hidalgo-Tobón, Silvia S., Dies-Suárez, Pilar, So, Po-Wah
• Format: Article
• Language: English
• Subjects: Abdominal subcutaneous adipose tissue; Body composition analysis; CNN; Fat quantification; MRI; Pediatric obesity; Visceral adipose tissue
• ISSN: 1746-8094
• DOI: 10.1016/j.bspc.2024.107250

•CNNs to quantify ASAT and VAT from MRI data automatically and accurately.•Crude volumes of ASAT and VAT were transformed to novel Total Intensity Maps.•Created Total Intensity Maps were novel matrix inputs into CNNs.•Fat quantifications were comparable to that obtained by the gold standard.•New readily accessible method without the need for high-power computing resources. Childhood obesity is a significant global health concern. Visceral adipose tissue (VAT) is more closely linked to the development of metabolic and cardiovascular diseases than abdominal subcutaneous adipose tissue (ASAT). This study aimed to develop a straightforward and automated method to quantify both ASAT and VAT in children from MRI. Dixon-based 3 T-MRI scans were conducted on Mexican boys aged 7–9 years. Novel Total Intensity Maps were generated from these scans for input into simple 2D Convolutional Neural Networks (CNNs) for automatic prediction/quantification of ASAT and VAT. The results were compared with the commercial semi-automated quantification and validated method provided by AMRA® Researcher which is the gold standard. The CNN-based method produced ASAT and VAT quantifications comparable to those obtained by AMRA® Researcher (ASAT, p = 0.1765; VAT, p = 0.7757), with a high correlation between the two methods for ASAT (R2 = 0.98, p