Article Text
PDF
Abstract
Background Extremely low weight and rapid changes in weight and body composition have become major concerns in many sports, but sufficiently accurate field methods for body composition assessment in athletes are missing. This study aimed to explore the use of ultrasound methods for assessment of body fat content in athletes.
Methods 19 female athletes (stature: 1.67(±0.06) m, weight: 59.6(±7.6) kg; age: 19.5(±3.3) years) were investigated by three observers using a novel ultrasound method for thickness measurement of uncompressed subcutaneous adipose tissue and of embedded structures. Two observers also measured skinfold thickness at eight International Society for the Advancement of Kinanthrometry (ISAK) sites; mean skinfold values were compared to mean subcutaneous adipose tissue thicknesses measured by ultrasound. Interobserver reliability of imaging and evaluation obtained by this ultrasound technique: intraclass correlation coefficient ICC=0.968 (95% CI 0.957 to 0.977); evaluation of given images: ICC=0.997 (0.993 to 0.999).
Results Skinfold compared to ultrasound thickness showed that compressibility of subcutaneous adipose tissue depends largely on the site and the person: regression slopes ranged from 0.61 (biceps) to 1.59 (thigh) and CIs were large. Limits of agreement ranged from 2.6 to 8.6 mm. Regression lines did not intercept the skinfold axis at zero because of the skin thickness being included in the skinfold. The four ISAK trunk sites caused ultrasound imaging problems in 13 of 152 sites (8 ISAK sites, 19 athletes).
Conclusions The ultrasound method allows measurement of uncompressed subcutaneous adipose tissue thickness with an accuracy of 0.1–0.5 mm, depending on the probe frequency. Compressibility of the skinfold depends on the anatomical site, and skin thickness varies by a factor of two. This inevitably limits the skinfold methods for body fat estimation. Ultrasound accuracy for subcutaneous adipose tissue measurement is limited by the plasticity of fat and furrowed tissue borders. Comparative US measurements show that skinfold measurements do not allow accurate assessment of subcutaneous adipose tissue thickness.
- Body composition methodology
- Eating disorders
- Weight loss
- Ultrasound
- Assessing validity and reliability of test of physiological parameters
Statistics from Altmetric.com
- Body composition methodology
- Eating disorders
- Weight loss
- Ultrasound
- Assessing validity and reliability of test of physiological parameters
Background
Improved field methods for body composition assessment in athletes are urgently needed
In sports medicine, extremely low weight, rapid changes in weight and body composition, and associated eating disorders with their severe health consequences have become a major concern. Severe restriction of food intake can result in nutrient deficiencies and chronic low body fat content may disrupt endocrine function. In weight-sensitive sports, team doctors and other healthcare professionals are faced with the difficult question of how to minimise health and performance risks for athletes.1 The current status of body composition assessment in sports has recently been reviewed by the International Olympic Committee (IOC) Working Group on Body Composition, Health and Performance,2 ,3 and heuristic suggestions on how to approach elite athletes who seek to achieve an unrealistic body composition and how to protect their health are discussed in a separate publication.4 In ‘weight-sensitive sports’ (gravitational sports, aesthetic sports and weight class sports),1 ,2 low weight may, at least in the short term, improve performance, but the possible effects have to be seen in the context of all other performance determinants. The health of the athlete is a precondition for optimum performance, but rapid and severe loss of tissue can cause disastrous performance setbacks and severe illness.5––23 Better protection of health and improved support of performance both depend on the availability of accurate and valid methods for the assessment of body composition.24 In ski jumping, for example, penalties (shorter skis) are applied to athletes whose body mass index (BMI) is deemed to be too low,1 ,25––28 but BMI is not a measure of body fat and is a very rough measure of ‘relative weight’.2
The IOC Working Group on Body Composition, Health and Performance has emphasised the need for development of improved body composition measurement techniques. These should be suitable for use in the field and should provide sufficiently accurate and reliable data. All the techniques in common use have major inherent problems, whether in methodology, interpretation of data or in the assumptions they make.1 ,2 ,24 ,29 ,30
In this study, a brightness-mode ultrasound (US) technique for subcutaneous adipose tissue (SAT) imaging without tissue compression combined with an evaluation software for multiple thickness measurements of SAT layers in athletes, which can be controlled visually, is introduced. The aim of this study is to propose this novel US technique, and to compare the measured uncompressed SAT to skinfold (SF) thicknesses.
B-mode US applied to thickness measurements in SAT
B-mode US images are generated by sequences of US beams, which are sent into the tissue in order to create an image in which the brightness of the screen corresponds to the echo intensity in the plane of the scan. The differences in acoustic impedance are quite large between the skin and adipose tissue and between the adipose tissue and muscle fascia; therefore, the transition from SAT to adjacent tissues can usually be clearly imaged when the US imaging parameters are chosen properly. This requires selection of anatomical sites where it is easy to make out skin, SAT and muscle fascia. The principle of US imaging is the pulse-echo technique. A short pulse is applied which travels at a speed of c (speed of sound) in the tissue. Diagnostic US systems conventionally use c=1540 m/s for calculating the distance (DUS) from the probe to the boundary between two tissues: DUS=cT/2, with T being the echo time. Diffraction and minimum pulse length limit lateral and axial resolution approximately to the wavelength used. Frequencies of US probes (transducers) applied in medical imaging range from 3 to 25 MHz, which corresponds to a wavelength range in soft tissue from 0.5 to 0.06 mm, and a typical investigable depth of 200 to 5 mm, respectively (attenuation of US increases with increasing frequency).31 ,32
US was used for estimation of fat thickness as early as 1965 by Bullen et al.33 In 1966, Booth et al34 published a comparison of US and Harpenden calipers. Since then, there has been growing interest in assessing body composition by means of US.35 Bellisari et al36 found that the measurement error in intraobserver and interobserver studies was less than 0.15 mm at all investigated sites (except for the triceps where it was 0.6 mm). One of the drawbacks in SF calliper measurements which can be overcome by US is the measurement uncertainty due to compressibility and viscoelasticity of adipose tissue.37 Non-invasive measurement of skin thickness using high-resolution US (above 20 MHz) has been developed as a preoperative evaluation tool in dermatology.38 ,39 Skin thickness does not change noticeably until the age of 60, diminishing thereafter.40 ,41 Results from highly accurate skin thickness measurements, for example, as published by Moore et al,40 can be helpful for thickness measurements of the SAT layer in cases in which the transition from skin to adipose tissue is difficult to identify. The mean skin thickness varies significantly among sites (ranging from about 1.0±0.15 mm in the upper arm to about 2.2±0.35 mm in the anterior abdomen). US has also been applied to the measurement of visceral fat. Koda et al42 compared US subcutaneous and visceral fat measurements with serial slice MRI and found that both subcutaneous and visceral fat measurements could be easily and reliably measured by US. They also found that US was superior to the calliper technique in measuring subcutaneous fat of obese persons.
Horn and Müller43 compared US measurements of SAT (DUS) in excised pig tissue with micrometer calliper (DMC) measurements (figure 1). For this comparison, a linear probe (7.5 MHz) was positioned with its centre at a marked position on the skin of the excised tissue and the US measurement of SAT thickness was made without compression, due to a thick layer of gel. Thickness of the adipose layer was evaluated in the vicinity of the centre of the US image.
Thickness measurements (DMC) of subcutaneous adipose tissue (SAT) in excised tissue from the pig. 140 DMC values obtained with a micrometer caliper with 0.01 mm accuracy were compared to ultrasound (US) B-mode image evaluations (DUS, 1510) at room temperature (using the edge detection algorithm for image evaluation). A slope of the regression line of 1.00, as well as a correlation coefficient of 0.998, resulted when c=1510 m·s−1 was used for distance calculation. Data from Horn and Müller.2 ,43
Underneath the marked site on the skin, SAT thickness was also measured in the slice of tissue by means of a micrometer calliper with a resolution of 0.01 mm. SE of the estimate (SEE) was 0.21 mm. The slope of the regression line was 0.98 when conventional sound velocity of c=1540 m/s was used (the regression coefficient reached 1.00 when c was set to 1510 m/s, indicating a lower speed of sound in adipose tissue). Statistical data are summarised in table 1.
Comparison of US (DUS, 1510) and micrometer calliper (DMC) fat measurement
Error of US and error of the micrometer calliper contributed equally to the deviations from the regression line (figure 1) because the plasticity of fat and the furrowed tissue transition zones naturally limit the accuracy of any measurement method to about ±0.2 mm or above, depending on the individual variability in the structure of the subcutaneous tissues.
The aim of the present study was to explore the use of the US method for assessment of SAT in athletes, as well as to compare the method with the widely used skinfold (SF) calliper method.
Methods
Subjects and measurement sites
Ethical approval
Permission to undertake the study was provided by the Ethics Commission of the Medical University of Graz (20–295ex08/09). All athletes received an information letter and completed a written consent form. The athletes and their coaches also had an opportunity to discuss the methods and aims of the study with the investigators. Written parental consent was required for participants below the age of 16 years.
Athletes
In order to span a sufficient range of SAT thicknesses, the group of athletes included 11 female football players (F) from the Austrian 2nd league and eight international and national level rhythmic gymnasts (G). Athletes’ data are given in table 2 (for further details see Müller et al44).
Athletes (test persons), age and anthropometric characteristics of female football players (F) and rhythmic gymnasts (G)
Observers
Three observers took US images of 19 athletes at each of the eight sites defined by the protocol of the International Society for the Advancement of Kinanthrometry (ISAK): The ISAK protocol defines eight sites: triceps, subscapular, biceps, iliac crest, supraspinale, abdominal, front thigh and medial calf.45–47 The sites had been marked on the skin by the first observer. Observers 1 and 2 were ISAK level 1 certified and they measured the SFs and all three observers measured the thickness of SAT by means of US. All three observers were briefly instructed on how to handle the US equipment for taking US images at the ISAK locations. None of the three observers had previous US experience.
Measurement sites
Three observers took US images (total number NUS=19×8×3=456) at the eight sites defined by the ISAK protocol and, in addition, two of them made SF measurements (NSF=19×8×2=304). The observers excluded 51 unclear images which could not be evaluated.
SF methods
Brief description of the ISAK recommended technique used here:45–47 approaching the skin surface at 90° with the finger and thumb; aligning these on the SF landmark; raising the SF with parallel sides; applying the calliper 1 cm laterally from the raised fold and midway along the fingernail of the index finger; releasing the calliper spring and reading the dial to 0.1 mm after two seconds. SF measurements were made immediately before the US measurements.
B-mode (US) imaging of SAT
Because the subcutaneous layer is compressible and shows viscoelastic behaviour, the US probe (transducer) was placed on a given site without any pressure by using a thick layer of US gel between the probe and the skin (typically 3–5 mm of the US gel should be seen as a dark band in the US image). Measurements were made in the standing position. The probe was held parallel to the direction of the SF. Conventional B-mode US systems were used (GE Logitec and Siemens AcusonX300PE; linear probes with 12 and 11.4 MHz, respectively—the according axial resolution was about 0.1–0.2 mm).
Semi-automatic thickness evaluations in B-mode US images of SAT
In 405 images, each observer defined the region of interest (ROI) in the vicinity of the central US ray (corresponding to the ISAK site) and made a series of thickness evaluations (semiautomatically). Sound speed c=1480 m/s was applied for distance calculation (in adipose tissue, c is lower than 1540 m/s, which is the value that is conventionally used).32 ,48 A sound speed deviation of 60 m/s would result in 4% thickness measurement error. For measurement of a series of thickness values, a region-growing algorithm for detecting SAT was developed. US images were exported and evaluated by means of the semi-automatic distance measurement algorithm. Within the region of interest, the software detects the contour of the adipose layer. The tissue segmentation was interactively (visually) controlled by changing a factor that determines accepted image inhomogeneity, and was conclusively stopped when the tissue segmentation matched the adipose tissue contour. The software automatically measures a series of distances in the region of interest, and minimum, maximum, mean, SD, median, mode and number of values are displayed. This region-growing software for SAT analysis also enabled the operator to distinguish between distance values in which other embedded tissues (eg, fibrous tissues or vessels) were included (DUS, incl) or excluded (DUS, excl). To compare measurement methods, selected thickness evaluations were also made using an edge-detection algorithm developed earlier (figure 2).2 ,43
B-mode ultrasound (US) image and subcutaneous adipose tissue (SAT) thickness evaluation (ISAK site: triceps) US probe (Siemens, 11.4 MHz) held parallel to skinfold (SF)-orientation and perpendicular to the skin. A thick layer of US gel, which ensures zero pressure, can be seen on top of the images (dark zone above the skin). Lower left: mean thickness of SAT (measurements in the region of interest: n=86) was 8.8 mm in this case, using the edge detection algorithm. Right side: Application of the region-growing algorithm resulted in distances of 8.9 mm and 7.7 mm (n=75), with fibrous tissues included or excluded, respectively. Image depth was 25 mm.
Statistics
Data of athletes are given as means±SD. Statistical analysis was performed with SPSS (IBM SPSS Statistics V.19) and GraphPad Prism 5. Box plot statistics were used for visualising distributions. For linear regression analysis, variables were pre-evaluated for normal distribution (Kolmogorov-Smirnov test, p>0.05). A p value <0.05 was considered as significant.
Regression analysis: slope, intercept and corresponding SDs, significance (p values) and 95% CI [conflow, confhigh] as well as Pearson’s regression coefficient-squared (R2) and SEE are presented. For statistical analysis, see Heyward and Wagner.49 Bland-Altman test: limit of agreement (LOA).50
Results
Figure 2 shows an example of a US image of SAT taken at the triceps site. Structures of relevance for US image evaluation are marked with black arrows, and tissue segmentation results by means of edge detection (left) and region-growing (right) algorithms are shown; mean values of SAT thicknesses in the region of interest were 8.8 (±0.4) mm (n=86) and 8.9 (±0.1) mm (n=75), respectively; the mean thickness without embedded fibrous structures was 7.7 (±0.1) mm (n=75). The seeds from which SAT detection started to grow are indicated as circles in the image on the right.
Figure 3 shows an unclear US image which led to erroneous evaluation (wrong choice of the region of interest) because there is a superficial layer of the abdominal wall (Camper's fascia) that contains a varying quantity of adipose tissue.
Ultrasound (US) image of subcutaneous adipose tissue (SAT) at the ISAK site abdominal. US probe (11.4 MHz) was held parallel to skinfold (SF)-orientation and perpendicular to the skin. Two white bands, one due to an intermediate fascia, led to the wrong region of interest (ROI) setting in the left image and thereby to a large distance difference when compared to the right image. Image depth: 40 mm.
Owing to the complex structure of the abdominal wall, evaluations of the three observers differed substantially in 10 cases. At the subscapular site, investigators erroneously evaluated unclear structures in nine cases. In order to get correct ratios of SF thickness to uncompressed SAT layer thickness measured by US, these 19 wrong evaluations (resulting in the elimination of 6 data-triplets) are not included in figure 4A (for analysis of interobserver differences, in a separate publication,44 data sets with and without these erroneous evaluations are used).
(A) Skinfold (SF) thickness (DSF) and ultrasound (US) measurements of subcutaneous adipose tissue (SAT) (DUS). For all US distance evaluations, a sound speed of 1480 m/s was used. Bold circles: US distances including embedded structures (DUS, incl); diamonds: US distances excluding embedded structures(DUS, excl). Displayed on top of each figure: ISAK sites at which US images were taken, R2-values for DUS, incl (p<0.05 at subscapular, all others: p<0.001), relations of DSF/DUS, incl (determined at DUS, incl=10 mm), number of data-triples (DSF, DUS, incl, DUS, excl) and R2-values for DUS, excl. All 130 data-triples from all 19 athletes and all eight ISAK sites are shown in the right figure, lower trace (from a total of 152 data-triples, 12 were eliminated because in 12 triples one or two US evaluations were missing, six triples were eliminated due to unclear images and three SF measurement mistakes (all in the front thigh) had to be corrected); regression line equations: DSF=1.251 DUS, incl+2.67; DSF=1.343 DUS, excl+3.17. (B) Percentage of fibrous and other structures embedded in SAT. The mean percentage of fibrous and other structures than fat detected in all 386 evaluated images from 19 athletes at eight ISAK sites was 12.6% (±9.6%), whereas the median was 10.19%.
At 13 sites (19 athletes, 8 sites each), none or only one of the three US images taken by the (inexperienced) investigators was clear enough for evaluation. In three cases, the SF measurement was substantially overestimated and clearly included muscle erroneously in the thigh SF.51 Therefore, the number of sites used in figure 4A is 130 (152-13-3-6): the 19 wrong image evaluations resulted in elimination of six sites.
In figure 4A, the mean US values of the three observers are plotted against means of SFs (two ISAK level 1 certified practitioners). Linear regression lines for both distances including fibrous structures (DUS,incl: bold circles) and excluding them (DUS,excl: diamonds) are drawn. Slopes of the regression lines (DUS, incl, bold solid lines) vary in a wide range, indicating different compressibilities of SAT at different sites. Slopes and intercept values of the regression lines obtained with data of the eight ISAK sites and statistical evaluations are shown in table 3. Regression lines did not intercept the DSF axis at 0 because at DUS, incl=0, the DSF value would include the two layers of (compressed) skin. Particularly in athletes with low fat, the contribution of the skin thickness is a substantial part of the SF thickness DSF, and therefore relations of DSF/DUS, incl, all taken at the same SAT value of DUS, incl=10.0 mm, are additionally given in the individual parts of figure 4A. In the bottom row, right figure, all data are combined, resulting in a slope value (for DUS, incl) of 1.29 and an intercept of 2.2; the intercept of the regression line for data DUS, excl (thicknesses without fibrous structures) was 2.8 mm. The mean of these intercepts (2.5 mm) indicates an average (compressed) skin thickness of about 1.25 mm (2.5/2; mean of all ISAK sites), but this is only a rough estimate due to the limited number of data and the methodical limitations of SF measurements.
Skinfold (SF) thickness DSF and ultrasound (US) measurement DUS, incl of SAT at ISAK sites: statistical data according to figure 4A.
Compared to US thickness, SF showed that compressibility of SAT depends largely on the site and person: slopes ranged from 0.61 (biceps) to 1.59 (thigh) and the CIs were large. Limits of agreement ranged from 2.6 to 8.6 mm. Regression lines did not intercept the SF axis at 0 because of the skin thickness included in SFs.
The mean percentage of structures other than fat within the SAT layer (386 evaluated US images) was 12.6% (± 9.6%). The amount of fibrous and other structures embedded in the adipose layer varies largely (figure 4B), depending on the site and person under investigation: the median was 10.2%, 50% of values were between 5.7% and 17.2%, and eight outliers and three extreme values were found.
Discussion
Interactive segmentation algorithms for adipose tissue detection
It is important for the investigator to interactively control the output of the SAT segmentation algorithm visually in order to ensure optimal segmentation and thereby prevent erroneous image interpretation: analysis and interpretation of complex biological structures still cannot be replaced in all cases by computer algorithms. The region-growing algorithm developed for adipose tissue detection also enables segmentation and measurement of embedded fibrous tissues.
Choice of US measurement sites
Sites for US measurement of subcutaneous adipose layer thickness should show simple structures; the thickness of the layer should not change appreciably in the vicinity, and their predictive value for total body fat or the fat content of body parts should be high. A systematic screening for optimal US sites all over the body surface has not yet been performed. The four ISAK sites on the limbs showed simple US structures; although the observers were not specialists in anatomy, no identification problems arose. However, at the trunk sites, the US observers were faced with anatomical identification problems; in particular at the abdomen, an intermediate fascia (Camper’s fascia) caused nine untypically large measurement deviations (between observers); they were not aware of this fascia and consultation of an anatomist whose research field deals with fasciae and embedded tissues in fat resulted in the advice to classify these results as erroneous due to anatomical misinterpretation. The supraspinale, iliac crest and abdomen sites accounted for 61% of images which could not be evaluated due to unclear structures (43 of a total of 70 unclear images). In such cases, visual observation of US image changes when athletes contract underlying muscles can be expected to reduce (or eliminate) erroneous US image interpretations. Interobserver variability analysis of the three US observers using both data sets (with explainable outlayers or without) is presented in a separate publication in this issue of the BJSM.44
A better choice of US sites on the trunk in which anatomical boundaries of adipose tissue are easily detectable (all four ISAK limb sites fulfilled this criterion) will be the best solution to keep such misinterpretations to a minimum. At the subscapular site, the presence of two visible bands in the US image instead of just one muscle-fascia boundary also led to image interpretation confusions in 14 cases (therefore, another 20% of the 70 images could not be evaluated correctly). All four limb sites together accounted for only 19% (13 images) of the non-evaluable images, that is, less than 3% of all 456 images could not be evaluated due to problems at one of the four limb sites. In figure 2, an example of a triceps measurement is shown in which the adipose tissue borders (dermis/adipose and adipose/muscle-fascia) are in the form of clearly visible continuous bands, and therefore evaluation is easy and error margins are small. Similarly simple structures were also found in the biceps, front thigh and medial calf.
US versus SF measurement
In the present study, the eight ISAK SF sites45––47 were used for two reasons: first, these sites are well known to many people as an international standard, and second, because comprehensive SF data of athletes already exist and US measurements allow the correlation between SF and uncompressed SAT to be assessed: this may improve the value of SFs that are used as predictors of total body fat.52––58 However, the predictive value is limited because of heterogeneity in SF compression at various sites and among individuals, and also because skin thickness varies between sites.40 ,59 Figure 4A shows SF values compared to US measurements of SAT at the ISAK sites. The range of significant relationships (R2) between SF measurements and uncompressed subcutaneous fat thicknesses [0.58, 0.88] was similar to the range obtained by radiographic fat measurement [0.6, 0.9],59 except for the value obtained in the study here at subscapular (0.34); the subscapular site does not seem to be a good choice for US measurements, at least until detailed specifications on how to proceed at this site are developed. Radiographic measurements showed that uncorrected SF measurements are not equivalent to uncompressed fat thickness in a particular site. At some sites in women, Himes et al59 found such large compression effects that the median single, uncompressed radiographic fat thickness was greater than the median SF thickness: the SF includes two layers of skin and two of SAT so this is only possible because of the high compressibility of the SAT layer when taking SFs. A comparable behaviour was also found here: the ratio (DSF/DUS), determined at DUS=10 mm, ranged from 0.9 (biceps) to 1.6 (at the iliac crest and medial calf), with an extreme value of 2.0 at the front thigh. At the only site which was identical in both studies, the medial calf, the ratios were the same (1.6). Stewart summarised that for an equation for total body fat assessment to have validity, five assumptions are to be met45: “constant compressibility, same skin thickness at all sites, constant fat fraction of adipose tissue, constant adipose tissue patterning and constant ratio of external to internal adipose tissue.” The validity of SF measurements may be improved by applying the US approach for investigations of compressibility, skin thickness (which is at the abdomen, for instance, approximately twice as thick as at the biceps),40 fraction of adipose tissue and tissue patterning. Visceral fat has also been investigated by means of US: when various methods were compared to CT, US was found to be the best alternative.42 ,60 ,61
Measurement of collagenous fibres embedded in SAT
The region-growing algorithm developed for this study is capable of identifying differing image structures and borders of embedded tissues. The content of other tissues than fat, such as elastic and collagenous fibres or vessels, varies largely in adipose tissue, as can be seen in figures 4A,B. Further anatomical and histological studies of these embedded structures and how they are represented in the US B-mode image is required to maximise the predictive value of US measurements for the estimation of total body fat.
Further applications
Owing to the accuracy that can be achieved with this US technique,2 ,43 ,44 comparative measurements may also be used for calibrating other imaging techniques like MRI or CT. Measurement parameter setting and image segmentation protocols, which are always crucial problems in MRI and CT (pixel size in MRI whole body scans is typically 2×2 mm, which limits the measurement accuracy accordingly), could be optimised this way.
Summary and conclusions
Subject to certain standardisations, US can offer a reliable estimation of the thickness of the subcutaneous adipose layer. The main advantages of the US technique for SAT measurement are: high spatial resolution, high measurement accuracy in the sub-mm range (provided the US image represents the site of interest), the visual control of the semiautomatic measurement algorithm (by which unavoidable errors of the automatic contour detection can be eliminated), the automatic measurement of many thickness values in one image, applicability in the field and the possibility to include or exclude embedded structures like fibrous tissues or vessels in the thickness values.
US enables fat patterning analyses. It might be possible in the future to determine the total volume of SAT by combining series of US measurements with body surface area measurement techniques such as three-dimensional photonic scanning. The accuracy that can be achieved for measurements of uncompressed SAT2 ,43 and high interobserver reliability44 supports the expectation that total body fat or total subcutaneous fat mass assessment based on US measurement should result in better estimates than other field methods. However, appropriate protocols and equations for US-based predictions have not yet been developed and sufficiently large screening studies in various groups of athletes are missing. As has been the case with establishing accurate SF measures (ISAK protocol),47 anatomical knowledge and experience are needed to identify correctly the interfaces of tissues in the US image, particularly when sites are used at which the anatomical structure is complex. The triceps, biceps, front thigh and medial calf sites were found to be easily evaluable, but reliable application protocols for US measurements on the trunk need to be developed. The predictive value of (uncompressed) tissue thickness data obtained by US will have to be analysed within the framework of validation studies using multicomponent body composition models.2 ,58 ,62––64 Interobserver reliability has also been analysed within the framework of this comparative study, and these results are the subject of a separate publication.44
What are the new findings?
-
The ultrasound (US) technique introduced here enables measurement of an uncompressed subcutaneous adipose tissue (SAT) thickness with an accuracy of 0.1–0.5 mm, depending on the probe frequency.
-
The method also allows the determination of the amount of fibrous structures embedded in SAT.
-
A comparison of US thickness measurements of SAT to skinfolds (SFs) showed that compressibility of SAT largely depends on the anatomical site; additionally, skin thickness varies by a factor of 2. This explains the accuracy limitations of SF methods for SAT thickness determination. SFs measure a compressed double layer of SAT and skin: when compared to the US technique, the correlation coefficients are low (0.34–0.88 in the group studied) and the limits of agreement are high (2.6–8.6 mm).
-
Not all International Society for the Advancement of Kinanthrometry (ISAK) sites used for SFs are good choices for US thickness measurements of SAT, because some cause anatomical identification problems.
How might it impact on clinical practice in the near future?
-
The results support the application of ultrasound (US) as a field (and laboratory) method for assessing subcutaneous adipose tissue in athletes with high accuracy and reliability (interobserver reliability is analysed in a separate publication in this issue of the BJSM).
-
Since this US method measures uncompressed SAT layers with high accuracy (which is limited only by biological reasons: furrowed borders and plasticity of SAT) and because high-resolution US devices can now be purchased at reasonable prices, there is a large potential of this new method to replace conventional field techniques like SFs or bio impedance analysis (BIA) with their well-known inherent shortcomings.
-
Owing to the obtainable accuracy, this US measurement technique may also be used for calibrating other SAT imaging techniques like MRI or CT.
Acknowledgments
The authors would like to thank the members of the IOC Ad Hoc Working Group on Body Composition, Health and Performance for many discussions preceding this study, F Anderhuber for anatomical support, W Gröschl and P Rohrer for US measurements and evaluations and K Pfeiffer for the thorough review of the manuscript.
References
Footnotes
-
Contributors WM designed the study and conducted it in co-operation with AF-R. MH developed the edge detection algorithm, and HA and PK developed the region-growing algorithm used for SAT analysis; JK and HA performed the statistical analysis, and RJM contributed to the discussion of the results and to the clearness of their presentation.
-
Competing interests HA, MH and WM who developed the US method applied here, are thinking of making the evaluation software commercially available.
-
Patient consent Obtained.
-
Provenance and peer review Not commissioned; externally peer reviewed.