Automatic measurement of human subcutaneous fat with ultrasound

Deep-learning based segmentation of ultrasound adipose image for liposuction

BACKGROUND

To develop an automatic and reliable ultrasonic visual system for robot- or computer-assisted liposuction, we examined the use of deep learning for the segmentation of adipose ultrasound images in clinical and educational settings.

METHODS

To segment adipose layers, it is proposed to use an Attention Skip-Convolutions ResU-Net (Attention SCResU-Net) consisting of SC residual blocks, attention gates and U-Net architecture. Transfer learning is utilised to compensate for the deficiency of clinical data. The Bama pig and clinical human adipose ultrasound image datasets are utilized, respectively.

RESULTS

The final model obtains a Dice of 99.06 ± 0.95% and an ASD of 0.19 ± 0.18 mm on clinical datasets, outperforming other methods. By fine-tuning the eight deepest layers, accurate and stable segmentation results are obtained.

CONCLUSIONS

The new deep-learning method achieves the accurate and automatic segmentation of adipose ultrasound images in real-time, thereby enhancing the safety of liposuction and enabling novice surgeons to better control the cannula.

Regression models for near-infrared measurement of subcutaneous adipose tissue thickness

Obesity is often associated with the risks of diabetes and cardiovascular disease, and there is a need to measure subcutaneous adipose tissue (SAT) thickness for acquiring the distribution of body fat. The present study aimed to develop and evaluate different model-based methods for SAT thickness measurement using an SATmeter developed in our laboratory. Near-infrared signals backscattered from the body surfaces from 40 subjects at 20 body sites each were recorded. Linear regression (LR) and support vector regression (SVR) models were established to predict SAT thickness on different body sites. The measurement accuracy was evaluated by ultrasound, and compared with results from a mechanical skinfold caliper (MSC) and a body composition balance monitor (BCBM). The results showed that both LR- and SVR-based measurement produced better accuracy than MSC and BCBM. It was also concluded that by using regression models specifically designed for certain parts of human body, higher measurement accuracy could be achieved than using a general model for the whole body. Our results demonstrated that the SATmeter is a feasible method, which can be applied at home and in the community due to its portability and convenience.

Automated Estimation of Melanocytic Skin Tumor Thickness by Ultrasonic Radiofrequency Data

OBJECTIVES

High-frequency (>20-MHz) ultrasound (US) is a noninvasive preoperative tool for assessment of melanocytic skin tumor thickness. Ultrasonic melanocytic skin tumor thickness estimation is not always easy and is related to the experience of the clinician. In this article, we present an automated thickness measurement method based on time-frequency analysis of US radiofrequency signals.

METHODS

The study was performed on 52 thin (≤1-mm) melanocytic skin tumors (46 melanocytic nevi and 6 melanomas). Radiofrequency signals were obtained with a single-element focused transducer (fundamental frequency, 22 MHz; bandwidth, 12-28 MHz). The radiofrequency data were analyzed in the time-frequency domain to make the tumor boundaries more noticeable. The thicknesses of the tumors were evaluated by 3 different metrics: histologically measured Breslow thickness, manually measured US thickness, and automatically measured US thickness.

RESULTS

The results showed a higher correlation coefficient between the automatically measured US thickness and Breslow thickness (r= 0.83; P< .0001) than the manually measured US thickness (r = 0.68; P < .0001). The sensitivity of the automated tumor thickness measurement algorithm was 96.55%, and the specificity was 78.26% compared with histologic measurement. The sensitivity of the manually measured US thickness was 75.86%, and the specificity was 73.91%.

CONCLUSIONS

The efficient automated tumor thickness measurement method developed could be used as a tool for preoperative assessment of melanocytic skin tumor thickness.

Measurement of subcutaneous adipose tissue thickness by near-infrared

Obesity is strongly associated with the risks of diabetes and cardiovascular disease, and there is a need to measure the subcutaneous adipose tissue (SAT) layer thickness and to understand the distribution of body fat. A device was designed to illuminate the body parts by near-infrared (NIR), measure the backscattered light, and predict the SAT layer thickness. The device was controlled by a single-chip microcontroller (SCM), and the thickness value was presented on a liquid crystal display (LCD). There were 30 subjects in this study, and the measurements were performed on 14 body parts for each subject. The paper investigated the impacts of pressure and skin colour on the measurement. Combining with principal component analysis (PCA) and support vector regression (SVR), the measurement accuracy of SAT layer thickness was 89.1 % with a mechanical caliper as reference. The measuring range was 5-11 mm. The study provides a non-invasive and low-cost technique to detect subcutaneous fat thickness, which is more accessible and affordable compared to other conventional techniques. The designed device can be used at home and in community.

Increase in subcutaneous adipose tissue and fat free mass in women with polycystic ovary syndrome is related to impaired insulin sensitivity

OBJECTIVES

The present study was performed to search whether subcutaneous and whole body adipose tissue increase and they relate to measures of insulin sensitivity in women with polycystic ovary syndrome (PCOS).

MATERIALS AND METHODS

52 women with PCOS and 53 healthy controls, all with similar age and body mass index participated in the study. A skinfold caliper device was used to measure biceps, triceps, subscapular and suprailiac skinfold thickness (SFT). Mid-upper arm circumference (MUAC) was measured using a tape measure. Body fat distributions were determined by bioelectrical impedance analysis. Insulin resistance score was computed with the HOMA formula. Plasma adiponectin was measured by EIA.

RESULTS

SFT in all defined areas, MUAC, total body and trunk fat free mass, and HOMA score were higher in women with PCOS compared with healthy women, while adiponectin level was significantly lower. SFT values correlated positively with HOMA score, and negatively with blood adiponectin level. Regression analysis indicated, SFT in triceps and supscapular areas, trunk fat mass, trunk fat ratio, fat free mass and trunk fat free mass values as the most powerful predictors of HOMA score.

CONCLUSIONS

The present study showed that SFT in different body regions and fat-free tissue mass are increased in women with PCOS, with a significant relation to impaired insulin sensitivity.

Measurement of the skin-liver capsule distance on ultrasound RF data for 1D transient elastography

Vibration-controlled transient elastography (VCTETM) technique is routinely used in clinical practice to assess non-invasively the liver stiffness which is correlated to hepatic fibrosis. Adequate use of the VCTETM probe requires the knowledge of the distance between the skin and the liver parenchyma. This paper compares two methods to estimate this distance using spatial variations of the spectral content of ultrasound radiofrequency (RF) lines, obtained from a probe consisting of a single element ultrasound transducer placed in front of the liver right lobe. Results on a database of 188 patients, including normal-weight and obese persons, show that the spectral variance can accurately discriminate the subcutaneous fat from the liver tissue. The proposed algorithm works in real-time and is suitable for VCTETM scanning protocol setup.