Developing a Deep Learning Approach for Automated Body Composition Prediction in Newborns Using Ultrasound Images
Open Access
- 3 December 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 206355-206366
Abstract
Objective: Measurements of human body composition such as fat mass (FM) and fat-free mass (FFM) are critical for studying malnutrition and the effects of nutritional interventions. This study introduces research toward a novel ultrasound scanning protocol combined with a deep learning analysis pipeline for predicting body composition. Methods: We analyzed a clinical dataset of 65 premature infants, consisting of ultrasound images from three anatomical locations (biceps, abdomen, and quadriceps), and ground truth FM and FFM from air displacement plethysmography (ADP). Our investigation focused on determining: (1) the optimal data processing methods for this application; (2) suitable baseline deep learning models for prediction to guide our learning strategy; and (3) the anatomical locations and image regions most predictive of FM and FFM. Results: We demonstrate that: (1) pre-processing techniques such as denoising, median filtering, and data augmentation enhance performance; (2) by employing a modified EfficientNet-B1 architecture, we achieve fully automatic body composition predictions from ultrasound images; (3) images obtained from combinations of biceps and quadriceps, as well as biceps, quadriceps, and abdomen scanning locations, resulted in mean absolute percent error (MAPE) values of 26.1% and 25.32%, respectively. Finally, sensitivity analysis shows that FM and FFM prediction are influenced by different body parts, as well as adipose and muscle tissue thickness. Conclusion: This study represents the first demonstration of deep learning for automated human body composition prediction from ultrasound images and lays a critical foundation for a novel ultrasound scanning and interpretation protocol to assess malnutrition.Keywords
Funding Information
- Google Research Scholar Program
- Boston College Schiller Institute for Integrated Science and Society
- National Institute of Child Health and Human Development (K23HD104000)
This publication has 38 references indexed in Scilit:
- Body composition assessment with limited field-of-view computed tomography: A semantic image extension perspectiveMedical Image Analysis, 2023
- Utility of muscle ultrasound in nutritional assessment of children with nephrotic syndromePediatric Nephrology, 2022
- The current use of ultrasound to measure skeletal muscle and its ability to predict clinical outcomes: a systematic reviewJournal of Cachexia, Sarcopenia and Muscle, 2022
- Recent advances and clinical applications of deep learning in medical image analysisMedical Image Analysis, 2022
- Fully automated body composition analysis in routine CT imaging using 3D semantic segmentation convolutional neural networksEuropean Radiology, 2020
- Can Ultrasound Measures of Muscle and Adipose Tissue Thickness Predict Body Composition of Premature Infants in the Neonatal Intensive Care Unit?Journal of Parenteral and Enteral Nutrition, 2020
- A survey on deep learning in medical image analysisMedical Image Analysis, 2017
- Body composition assessment in the infantAmerican Journal of Human Biology, 2014
- Distinctive Image Features from Scale-Invariant KeypointsInternational Journal of Computer Vision, 2004
- Backpropagation and stochastic gradient descent methodNeurocomputing, 1993