Obesity is a major public health problem, the prevalence of which has increased worldwide [
Several methods for estimation of the visceral fat area (VFA) have been reported. Until now, computed tomography (CT) is considered as the gold standard method for the measurement of VFA [
The purpose of this study was to develop two novel US methods for estimation of the abdominal VFA. Then, multiple regression analysis was performed using several physiological parameters as covariates in order to identify the parameters that would significantly enhance the correlation with the VFA calculated by CT.
For easier and more accurate estimation of the VFA, we devised two novel US methods: the triangle method and the ellipse method. In the first, the triangle method, the VFA is assessed as a summed area of six triangles, and in the second, the ellipse method, the VFA is assessed as part of an ellipse. The VFA calculated by US was compared with the VFA calculated by CT.
Since a US probe cannot be placed easily on the surface of the umbilicus, and the aorta is normally located to the left of the center, a point 2 cm to the left of the umbilicus was used as the basal point. Two points, each 5 cm to the left and right of the basal point, were also added, and the measurements were conducted with the US probe placed at these three points. USdetermined visceral fat distance was defined as the distance between the internal surface of the rectus abdominis muscle and the posterior wall of the aorta from each diagnostic position (Figure
(a) US measurement picture; distance A showed the distance from the skin to the posterior wall of the aorta. Distance A1 showed the distance from the internal surface of the rectus abdominis muscle to the posterior wall of the aorta. The thickness of the subcutaneous fat layer showed the distance from the skin to the rectus abdominis muscle. (b) Triangle method; VFA was calculated as a summed area of the six triangles, which is calculated by the distance to the back wall of the aorta from three positions. (c) Ellipse method; we hypothesized that the ellipsoidalshaped peritoneal cavity was the reduced scale model of the ellipsoidalshaped cross section of abdomen. We defined waist circumference as the circumference of the ellipse, and the USmeasured distance as the semiminor axis. The VFA was taken by subtracting the back side onethird area occupied by bone and muscles from the peritoneal cavity ellipse.
US measurement
Triangle method
Ellipse method
We designed a beltshaped ultrasound probecompatible device to provide a quick, easytooperate, and accurate way to guide the ultrasonographic procedures in the triangle method (Supplementary Information available online at
Based on the recognition that the peritoneal cavity is ellipsoidal in shape, we hypothesized that the ellipsoidalshaped peritoneal cavity was the reduced scale model of the ellipsoidalshaped cross section of the abdomen. We defined waist circumference as the circumference of the cross section of the abdomen and the semiminor axis as the USmeasured distance from the skin to the posterior wall of the aorta (Figure
The circumference of the peritoneal cavity ellipse was calculated from the waist circumference and the ratio of the measured distances. The distance from the internal surface of the rectus abdominis muscle to the posterior wall of the aorta (distance A1) and the distance from the skin to the posterior wall of the aorta (distance A) were used for the calculation. The area of the peritoneal cavity ellipse was calculated based on the distance from the internal surface of the rectus abdominis muscle to the posterior wall of the aorta as the semiminor axis and the calculated circumference.
The back side onethird area was occupied by bone and muscles; therefore, this area was subtracted from the area of the peritoneal cavity ellipse. Then, the area of the intestinal tract (10 cm^{2}) was subtracted, to finally calculate the VFA.
To assess the intrarater and interrater reliability of the US measurements, we carried out a preliminary study on other subjects than the study subjects. This trial was carried out by two highly skilled sonographers in three male volunteers. The measurements were carried out 4 times per day on each man, on two different days. The interclass correlation coefficient (ICC) of the intrarater reliability was calculated by US measurements carried out several times in a subject by an expert US technician (Table
Intrarater reliability and interrater reliability were assessed by measuring the distances A1 and A shown in Figure
Intrarater reliability  Interrater reliability  

The distance A1  Investigator A  Investigator B  The distance A1  Correlation coefficient 
ICC (1,1)  0.9817  0.9970  ICC (2,1)  0.9988 
ICC (1,3)  0.9954  0.9992  ICC (2,3)  0.9987 
The distance A  Investigator A  Investigator B  The distance A  Correlation coefficient 
ICC (1,1)  0.9800  0.9971  ICC (2,1)  0.9988 
ICC (1,3)  0.9949  0.9993  ICC (2,3)  0.9987 
ICC: interclass correlation coefficient.
We recruited 100 volunteer males
In this study, four risk factors (high blood pressure, high triglyceride, low highdensity lipoprotein (HDL) cholesterol, and hyperglycemia) defined in the criteria of the National Cholesterol Education Program’s Adult Treatment Panel III guidelines in 2005 [
VFA was calculated by the two ultrasonographic (EUB8500, Hitachi Ltd., Tokyo, Japan) methods, the triangle method and the ellipse method. The following parameters were measured with a 3.5 MHz convex array probe: (1) the distance from the skin to the posterior wall of the aorta, (2) the thickness of the subcutaneous fat layer, and (3) the distance between the internal surface of the rectus abdominis muscle and the posterior wall of the aorta. Imaging was performed at the end of a normal expiration in the supine position. The US probe was placed against the skin as lightly as possible to prevent compression of the fat layers. The time required for US measurement was within one or two minutes. All the US measurements were carried out in duplicate by the same investigator, an expert US technician.
CT equipment from Toshiba Medical Systems (Tokyo, Japan) was used for the abdominal CT. Imaging was carried out at the end of expiration at the level of the umbilicus in the supine position. The scan interval was set at 7.5 mm. Standard and appropriate measurement methods for waist circumference are different by the country and race. We measured the waist circumference by the level of the umbilicus in accordance with the Japanese criteria of metabolic syndrome [
The CT images were analyzed using the Fat Scan ver.4 software (East Japan Institute of Technology Co., Ltd., Hitachi, Ibaraki, Japan) to calculate the abdominal VFA.
The study protocol conformed to the ethical guidelines of the 1975 Helsinki Declaration and was conducted with the approval of the ethics committee of each of the University of Tokyo, Hitachi Ltd., and Hiraka General Hospital. Written informed consent was obtained from each subject for participation in the study.
Pearson’s correlation coefficients were calculated to assess the association among the clinical parameters and the VFA calculated by CT. The statistical significance of differences in the continuous data between groups was examined by ANOVA. Statistical significance was set at
To enhance the accuracy of estimation of the VFA, multiple regression analysis was conducted. The parameters used for the analysis along with the new method were the height, weight, BMI, age, and waist circumference. It was needed that the data, which is applied regression method, should be normally distributed. So, as the first step, we confirmed whether the objective variable that is VFA is normally distributed by using the KolmogorovSmirnov test. Since the
The characteristics of the 100 male study participants are shown in Table
Characteristics of the study participants (
Average  SD  Minimum  Maximum  

Age (yr.)  39.6  11.0  22.0  63.0 
Waist 
84.4  9.2  63.0  106.4 
Height (cm)  172.1  5.3  158.1  181.7 
Weight (kg)  70.0  11.0  47.9  98.0 
BMI  23.6  3.3  16.9  34.2 
VFA (cm^{2})  79.5  39.0  9.8  198.7 
SFA (cm^{2})  141.0  66.8  17.0  337.0 
Total adipose 
220.4  94.3  32.3  489.3 
SBP (mmHg)  129.5  14.3  104.0  174.0 
DBP (mmHg)  80.1  12.0  58.0  119.0 
FBS (mg/dL)  90.1  11.6  65.0  135.0 
Insulin ( 
6.5  4.1  1.0  23.0 
HOMAR  1.5  1.0  0.2  5.9 
HbA1c (%)  5.7  0.3  5.1  6.9 
TC (mg/dL)  200.7  32.1  135.0  287.0 
HDLC (mg/dL)  61.7  18.3  30.5  142.2 
LDLC (mg/dL)  121.6  29.3  69.0  207.0 
TG (mg/dL)  156.8  140.6  15.0  744.0 
UA (mg/dL)  6.2  1.4  2.6  10.3 
BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; VFA: visceral fat area; SFA: subcutaneous fat area; FBS: fasting blood sugar; HOMAR: homoeostatic model assessment ratio; JDS: Japan Diabetes Society; TC: total cholesterol; HDL: highdensity lipoprotein; LDL: lowdensity lipoprotein; TG: triglyceride; UA: uric acid.
The correlations between each parameter of the subjects and the VFA calculated by US are shown in Figures
The correlations between the VFA calculated by US method and each of the parameters are shown. (a) Triangle method and waist circumference (
We next assessed the correlation between each parameter and the 4 risk factors for metabolic syndrome, namely, hypertension, hyperglycemia, and dyslipidemia. The waist circumference was significantly increased in the males with ≧2 risk factors as compared to that in the males without any risk factors (Figure
We used multiple regression analysis to assess the associations between the VFA calculated by CT and several physiological parameters. The results of multiple regression analysis carried out with the triangle method using the height, weight, age, waist circumference, and BMI are shown in Figure
(a) Multiple regression analysis with the triangle method (
Multiple regression analysis using various parameters.
Containing triangle method.
Triangle method  Height  Weight  Age  Waist circumference  BMl  Correlation coefficient 

○  ×  ○  ○  ○  ×  0.8586435 
○  ×  ○  ○  ×  ×  0.8582122 
○  ×  ×  ○  ×  ○  0.8565199 
○  ○  ×  ○  ×  ○  0.8559084 
○  ×  ○  ○  ×  ○  0.8555399 
○  ○  ○  ○  ×  ×  0.8551473 
○  ×  ×  ○  ○  ○  0.8547307 
○  ○  ×  ○  ○  ○  0.8544964 
○  ○  ○  ○  ×  ○  0.8544280 
○  ×  ×  ○  ○  ×  0.8543381 
○  ×  ○  ○  ○  ○  0.8540807 
○  ○  ○  ○  ○  ×  0.8540032 
○  ○  ○  ○  ○  ○  0.8528104 
○  ○  ×  ○  ○  ×  0.8500605 
○  ○  ×  ×  ○  ×  0.8073538 
○  ×  ×  ×  ○  ×  0.8048660 
○  ×  ×  ○  ×  ×  0.8029817 
○  ○  ×  ×  ○  ○  0.8026849 
○  ×  ○  ×  ○  ○  0.8026364 
○  ○  ○  ×  ○  ×  0.8025610 
○  ×  ○  ×  ○  ×  0.8019843 
○  ×  ×  ×  ×  ○  0.8018995 
○  ×  ×  ×  ○  ○  0.8017493 
○  ○  ○  ×  ○  ○  0.8006840 
○  ○  ×  ○  ×  ×  0.7993090 
○  ×  ○  ×  ×  ○  0.7984585 
○  ○  ×  ×  ×  ○  0.7982493 
○  ○  ○  ×  ×  ×  0.7974290 
○  ○  ○  ×  ×  ○  0.7969801 
○  ×  ○  ×  ×  ×  0.7878478 
○  ×  ×  ×  ×  ×  0.7586674 
○  ○  ×  ×  ×  ×  0.7512694 
Containing ellipse method.
Ellipse method  Height  Weight  Age  Waist circumference  SMI  Correlation coefficient 

○ 


○ 

○  0.8642723 
○ 

○  ○ 


0.8641731 
○  ○ 

○ 

○  0.8633550 
○ 

○  ○ 

○  0.8630552 
○  ○  ○  ○ 


0.8630401 
○ 

○  ○  ○ 

0.8619440 
○  ○  ○  ○ 

○  0.8617836 
○ 


○  ○  ○  0.8600242 
○  ○ 

○  ○  ○  0.8598155 
○  ○  ○  ○  ○ 

0.8595745 
○ 

○  ○  ○  ○  0.8594486 
○  ○  ○  ○  ○  ○  0.8581494 
○ 


○  ○ 

0.8564621 
○  ○ 

○  ○ 

0.8529285 
○ 




○  0.8162489 
○  ○ 


○ 

0.8158341 
○  ○ 



○  0.8134919 
○ 

○ 


○  0.8134300 
○  ○  ○ 



0.8130791 
○ 



○ 

0.8122610 
○  ○  ○ 


○  0.8122025 
○ 



○  ○  0.8118373 
○ 


○ 


0.8113445 
○  ○  ○ 

○ 

0.8109480 
○  ○ 


○  ○  0.8108606 
○ 

○ 

○  ○  0.8106689 
○  ○  ○ 

○  ○  0.8093496 
○  ○ 

○ 


0.8076415 
○ 

○ 

○ 

0.8066263 
○ 

○ 



0.8010528 
○ 





0.7731967 
○  ○ 




0.7677818 
○: model containing the parameter; ×: model not containing the parameter.
We carried out further multiple regression analysis to investigate which combination would show the best correlation with the VFA calculated by CT (Tables
To confirm the reliability of VFA estimation by using triangle method and ellipse method, we applied paired
We assessed the correlation between the results of the multiple regression analysis and the 4 risk factors for metabolic syndrome. The VFA calculated using multiple regression analysis with the triangle method was significantly higher in the men with any risk factors than in those with no risk factors (Figure
In this study, we describe two novel US methods for estimation of the abdominal VFA. The VFA calculated by the triangle method as well as that determined by the ellipse method showed a high correlation coefficient with the VFA calculated by CT. The VFA calculated by each of these methods was significantly increased in the men with one or more metabolic risk factors than in those with no risk factors. In addition, VFA calculated by these two methods with multiple regression analysis carried out using several parameters as covariates showed a higher correlation coefficient with the VFA calculated by CT.
The triangle method and the ellipse were revealed to be easy and accurate practical evaluation methods for the assessment of VFA. Although both the VFA values calculated by the triangle method and the ellipse method showed high correlation coefficients with the VFA calculated by CT, each of these methods has its own advantages and disadvantages. The advantage of the triangle method is that the calculation needs only US data without measurement of other parameters such as the waist circumference, although 3 points of measurement are needed, with a high level of skill in the US technique. Furthermore, the operation is slightly complicated owing to the use of the beltshaped device, and timeconsuming. Meanwhile, the ellipse method needs measurement of the waist circumference and the operation time is shorter because only one point of measurement by US is needed. Especially, detection of the internal surface of the rectus abdominis muscle and the posterior wall of aorta from 2 cm to the left side of the umbilicus (basal point) is comparatively easy by US, even in obese subjects. In fact, the intrarater and interrater reliability of US measurement at the basal point were very high, and the intrarater reliability was higher for measurement at the basal point than that at a distance of 5 cm from the basal point. (Table
In the multiple regression analysis, the age was an important factor for both the triangle method and the ellipse method to enhance the correlation coefficient with the VFA calculated by CT. Muscle mass and basal metabolic rate are known to decrease with aging, and increase of the body fat percentage without body weight change is known to be common in the elderly. Actually, some crosssectional studies have reported increased visceral fat in the elderly as compared with that in young people for the same BMI [
In regard to other VFA calculation methods, the VFA calculated by the bioelectrical impedance analysis showed a correlation coefficient
This study had the following limitations. First, the study participants were relatively young male volunteers, and therefore, whether the methods are suitable for other age groups remains unknown. Second, the sample size of this study was small, because it was a pilot study. Further large studies are required to assess the suitability of the triangle and ellipse methods for women, elderly people, and patients with metabolic syndrome. Third, whether the intrarater and interrater reliability would still be maintained with a higher number of examiners is unknown.
In conclusion, we devised fast and accurate ultrasonographic methods for the measurement of VFA. The VFA measured by the triangle method as well as that measured by the ellipse method showed a high correlation coefficient with the VFA calculated by CT. These US methods are easy to use, noninvasive, and do not involve radiation exposure, and the measurements can be carried out frequently. We hope that our simple method would be widely adopted for the evaluation of VFA.
Takeharu Asano is the first author.
Naoto Kubota has received research grants from Astellas Pharma. Inc. and speaker honoraria from Toray Industries Inc. and Astellas Pharma. Inc. Takashi Kadowaki has received commitment from Hitachi Ltd. as a part of TSBMI, which is a matching program for innovation based on an agreement between Tokyo University and Hitachi Ltd. Takashi Kadowaki has also received research grants from Astellas Pharma. Inc. and speaker honoraria from Toray Industries Inc. and Astellas Pharma. Inc. Takeharu Asano, Norihiro Koizumi, Kazuhito Yuhashi, Hongen Liao, Mamoru Mitsuishi, Shigemi Takeishi, Toshiaki Takahashi, Shin Ohnishi, Shiro Sasaki, Ichiro Sakuma have no conflict of interest to disclose.
Takeharu Asano and Naoto Kubota designed the study, performed the analyses, and helped in the manuscript writing. Shigemi Takeishi, Toshiaki Takahashi, and Shiro Sasaki collected the data. Kazunori Itani, Tsuyoshi Mitake, and Kazuhito Yuhashi performed the analyses and commented on the drafts of the manuscript. Norihiro Koizumi, Hongen Liao, and Ichiro Sakuma contributed to the interpretation and discussion of the results. Mamoru Mitsuishi, Shin Ohnishi, and Toshiaki Takahashi conducted the study. All authors have approved the final article.
The authors wish to express our sincere appreciation to Dr. Hiroyuki Tsukihara (Graduate School of Medicine, University of Tokyo), You Zhou (Graduate School of Engineering, University of Tokyo), and Yoko Fujihara (Hitachi, Ltd.) for their contribution to the data analysis and preparation of the manuscript. This work was supported by a grant for TSBMI from the Ministry of Education, Culture, Sports, Science and Technology of Japan, grantsinaid for Scientific Research in Priority Areas (A) (18209033) and (S) (20229008) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to Takashi Kadowaki).