Professor Hanyang University Seoul, Seoul-t'ukpyolsi, Republic of Korea
Background: Central obesity in children is a potent risk factor for cardiovascular diseases, even more significant than BMI-defined obesity, and it can persist into adolescence and adulthood. Especially apple-shaped central obesity, which means a relatively large waist circumference compared to hip circumference, is related to the deposition of a large amount of visceral fat. Measuring the Waist-to-Hip Ratio (WHR), an indicator of body fat distribution, typically requires direct measurement using a tape-based method, which can be inconvenient and socially unacceptable. Therefore, non-contact anthropometric techniques to measure WHR are necessary for routine assessment in clinical settings. Objective: This study proposes a non-contact approach for measuring WHR and analyzing body type using frequency modulated continuous wave (FMCW) radar. Design/Methods: In this study, a total of 65 participants (53% boys; mean age ± SD 10.7 ± 1.6 years) who visited the pediatric endocrinology outpatient clinic at Hanyang University Hospital in Seoul, South Korea, were enrolled. Data were collected using FMCW radar while participants stood upright for 5 seconds. For reference, the waist and hip circumferences were measured by a medical specialist. Images were generated from the data collected by the radar, and WHR was measured using a convolutional neural network (CNN). Based on the measured WHR values, participants were categorized into three body types: 'Apple' body type for WHR ≥ 0.9 for males and ≥ 0.85 for females, 'Pear' body type for WHR < 0.85 for males and < 0.80 for females, and 'Avocado' body type for WHR > 0.85 and ≤ 0.9 for males and > 0.80 and ≤ 0.85 for females. Results: For the WHR measured based on radar data, the Intra-class correlation coefficient (ICC) was 0.87, with a p-value less than 0.05, confirming a significant correlation between the actual body measurements and those obtained through radar technology. The accuracy, sensitivity, and specificity between the categorized body types based on the measured WHR and the actual body types were found to be 0.83, 0.85, and 0.92 respectively, demonstrating a high level of correlation.
Conclusion(s): The experimental results confirmed a high correlation between the WHR measured by radar and the actual measurements and the effective identification of body types such as 'Apple,' 'Pear,' and 'Avocado' through radar-based body type analysis. In future research, we will recruit more participants and check the anthropometric characteristics of participants in each group.
