Validity of a Structure Sensor-based Anthropometric Measurement:

Performing a Pilot Study

Darko Katović

, Marjeta Mišigoj-Duraković

and Anita Bušić

Faculty of Kinesiology, University of Zagreb, Horvaćanski Zavoj 15, Zagreb, Croatia

Live Good d.o.o., Zagreb, Croatia

Keywords: Anthropometry, Kinanthropometry, Reliability, Validity, 3D Body Scaning, Structure Sensor.

Abstract: Development of new technologies is offering possibilities to overcome “traditional” limitations of

anthropometric measures and enable the production of a new generation of simple, high-speed, inexpensive,

highly defined and precise scanners for superficial body imaging. This study is an attempt to determine the

metric characteristic of the instrument (BodyRecog PRO) which technology is based on the method of deep

infrared 3D-scanning (Structure sensor). Reliability of the digitally obtained anthropometric measures was

tested in the process of relating them with the measures obtained via the traditional anthropometric

quantification.

1 INTRODUCTION

A contemporary level of technological development

makes it possible to construct instruments that are

portable and mobile enough to meet requirements of

versatile scientific branches in the field of data

acquisition.

In this paper, the focus is on the need to acquire

anthropometric measures. The intense development

of technology in a couple of the last decades has

gradually upgraded a “traditional” model of

anthropometric measures (length, width, skinfolds’

thickness, circumferences) with more complex

measures such as volumes and surface sizes of the

measured objects, by which changes in body size and

shape can, relatively inexpensively, be detected in

real time quite precisely, the goal hardly achievable

by the traditional measuring instruments (Rønnestad,

Hansen & Raastad, 2010; Schranz et al., 2010, 2012).

The application of the already well-known and

generally accepted techniques of digital body

measuring, founded upon the three dimensional (3D)

systems for superficial imaging, is limited due to their

high purchase costs, complex implementation and

constrained accessibility. However, the development

of new technologies (sensors and cameras of high

definition, data processing using machine learning

and artificial intelligence…) is offering posibilities to

overcome the mentioned limitations; it has enabled

the production of a new generation of simple, high-

speed, inexpensive, highly defined and precise

scanners for superficial body imaging (Simmons &

Istook, 2003; Zhang et al., 2014; Ryder & Ball, 2012;

Bragança et al.).

Different types of body scan sensing based

technologies can be found at the market. They differ

regarding ease of use, and quality of 3D models

reconstructed.

Applicable value of such technologically

saturated instruments depends on various factors that

have a direct influence on reliability and validity of

the process of measurement.

This paper investigates the utility of specific 3D

body scan technology in relation to classical

anthropometric approach.

2 METHODS

Besides the classical antropometric instruments, a

newly constructed measuring instrument assessing

girths of body segments – BodyRecog PRO has been

used in the research. The objective was to test metric

characteristics of the instrument founded upon the

method of a infrared depth-sensing 3D-scanning

technology. Therefore, reliability of the digitally

obtaine anthropometric measures was tested in the

process of relating them with the measures obtained

via the traditional anthropometric quantification, thus

also testing validity of the new instrument.

Katovi

c, D., Mišigoj-Durakovi

c, M. and Buši

c, A.

Validity of a Structure Sensor-based Anthropometric Measurement: Performing a Pilot Study.

DOI: 10.5220/0008506902450250

In Proceedings of the 7th International Conference on Sport Sciences Research and Technology Support (icSPORTS 2019), pages 245-250

ISBN: 978-989-758-383-4

245

2.1 Digital Mesurement Technology

The digital measurement was conducted by an iPad

Air 2 tablet with the iOS (ver. 10) operational system

and a 3D-scanner Structure Sensor - model: ST01

(Occipital, 2019) and with support of BodyRecog

PRO software (ver. 0.9.19).

Structure Sensor’s technical specification

includes the following technical elements important

for the study: Maximum recommended range

(3.5m+), minimum recommended range (40cm),

precision (0.5mm at 40cm, 30mm at 3m), field of

view (Horizontal: 58 degrees, Vertical: 45 degrees),

Resolution (VGA 640x480, QVGA 320x240).

Each of the above mentioned has a direct impact

on the result obtained.

2.2 Subjects

A convenience sample consisted of 71 participants of

both genders (men: n=52, age in years: mean±21,7;

SD±0,7; women: n=19; age in years: mean±20,9;

SD±0,4), students of the Faculty of Knesiology in

Zagreb. Participants were first manually measured –

by a set of standard anthropometric measurements of

circumferences, and then a digital measurement was

conducted using a newly constructed measuring

instrument BodyRecog PRO.

2.3 Variables

Entities were described by the sample of 34

variables

, out of which 18 were variables obtained by

the classical measurement of anthropometric

dimensions, whereas 16 variables were obtained by

the digital measurement of anthropometric

dimensions; the latter variables were defined by the

body sites and points that were either equivalent to

the ones of the traditional anthropometric

measurement (three measurement trials), or were

repositioned in an acceptable way.

A group of the traditional anthropometric

measures consisted of the following variables: body

height (BH), body mass (BM), waist circumference,

abdominal circumference, hip circumference, neck

circumference, breast circumference, chest cavity

circumference, left upperarm circumference, right

upperarm circumference, left forearm circumference,

right forearm circumference, left wrist circumference,

right wrist circumference, left thigh circumference,

Measurement units (classical and digital measurement):

BodyMass (kg), BodyHeight (cm), Girths (cm), Diameters

(cm).

right thigh circumference, left lowerleg (calf)

circumference, right lowerleg (calf) circumference.

A group of the digital anthropometric measures

consisted of the following variables

: D-BodyHeight,

D-NeckGirth, D-WaistGirth, D-AbdominalGirth, D-

HipsGirth, D-ChestGirth, D-BreastGirth, D-

RightUpperArmGirth, D-LeftUpperArmGirth, D-

RightForearmGirth, D-LeftForearmGirth, D-

RightWristGirth, D-LeftWristGirth, D-

RightWristDiameter, D-LeftWristDiameter, D-

RightUpperLegGirth, D-LeftUpperLegGirth, D-

RightLowerLegGirth, D-LeftLowerLegGirth.

2.4 Measurement Protocols

The traditional anthropometric measurement used the

standard procedure, conducted according the

International Biological Programme (IBP) and using

the standard measurement instruments but for a slight

modification – the examinee’s position was adjusted

to the position assumed in the digital scanning (the

feet hip-width apart and the extended arms raised

laterally at the shoulder height). Extremity

circumference measurements were executed on both

sides.

The digital anthropometric measurement

followed the traditional one. For the standardisation

purposes, the digital anthropometric measurement

protocol was designed. Here is a shortened version:

The space within which measurement scanning is

conducted must be at least 3 x 3 m with the central

marker for the participant. The examinee stands

quietly with the feet hip-width apart facing the

measurer. The arms are in side raise, paralell with the

floor, with the palms facing the floor. The

participant’s gaze is directed straight forward

throughout the measurement procedure. The

measurer, facing the participant 2-2.5 m apart

(distance in calibration phase) and holding the iPad

with the scanner perpendicular to the floor and at the

height corresponding to the participant’s abdomen,

positions the reference framework of the software

(guided by the software). Upon the software signal

saying that the action has been executed properly, the

measurer circles around the examinee 1 m apart

(distance in digital scaning phase); iPad must be

perpendicular to the floor all the time and at the half

of the participant’s height. The measurer stops

circling for a while after every circle quarter in order

to enhance body contours’ imaging. The measurement

Prefix “D” denotes a digital measurement.

K-BioS 2019 - Special Session on Kinesiology in Sport and Medicine: from Biomechanics to Sociodynamics

246

is over after three successfully registered/recorded

repetitions.

2.5 Statistical Analysis

Data were processed using the statistical package

StatisticaDell Inc. (Dell, 2017). The used procedures

included the computation of descriptive parameters

(mean, standard deviation, total range, variability

coefficient, distribution form parameters: skewness

and kurtosis). Reliability, based on the traditional

anthropometric measurement model, was assessed

using the method of internal consistency to establish

the following reliability coefficients: Cronbach and

Spearman-Brown’s (standardised) alpha. Pearson’s

correlation coefficient was applied to determine a

diagnostic validity of the newly constructed measuring

instrument (BodyRecog PRO).

3 RESULTS AND DISCUSSION

Descriptive parameters of both the traditionally and

digitally measured circumference variables were

computed for each of the convenience subsamples of

female and male students.

Basic descriptive parameters (central – arithmetic

mean, dispersive – range and standard deviation) of the

traditionally measured variables are presented in

Tables 1 and 2.

Table 1: Descriptive parameters (clasical measuremet) –

male students.

Variable name Mean Ran

e Std.Dev.

Bod

Mass 80,346 41,900 8,537

BodyHeight 183,138 34,200 7,455

NeckGirth 39,048 6,300 1,442

WaistGirth 79,950 17,600 4,081

AbdominalGirth 82,627 20,300 4,897

sGirth 100,921 20,400 4,471

ChestGirth 98,950 24,000 4,607

BreastGirth 95,867 23,400 4,308

RightUpperArmGirth 31,548 9,200 2,331

LeftU

erArmGirth 31,244 10,600 2,431

htForearmGirth 28,469 8,000 1,440

LeftForearmGirth 28,062 7,300 1,479

RightWristGirth 17,677 3,800 0,765

LeftWristGirth 17,510 3,400 0,708

RightWristDiamete

5,946 1,400 0,298

LeftWristDiamete

5,871 1,100 0,255

htU

erLe

Girth 57,085 15,200 3,152

LeftU

erLe

Girth 56,692 16,100 3,189

RightLowerLegGirth 38,554 7,200 1,734

LeftLowerLegGirth 38,285 8,000 1,802

Values of the traditionally measured variables and

their parameters were in line with the values obtained

in the many same or similar previous measurements

conducted with the population of female and male

students of the Faculty of Kinesiology in Zagreb.

Table 2: Descriptive parameters (clasical measurement) –

female students.

Variable name Mean Ran

e Std.Dev.

Bod

Mass 62,000 28,300 6,929

BodyHeight 168,668 20,500 5,783

NeckGirth 32,837 5,900 1,408

WaistGirth 69,921 15,700 4,126

AbdominalGirth 77,363 15,700 5,032

sGirth 97,979 15,500 4,160

ChestGirth 86,174 16,200 3,583

BreastGirth 88,247 17,800 4,456

RightUpperArmGirth 27,132 7,000 1,778

LeftU

erArmGirth 26,595 6,600 1,713

htForearmGirth 24,068 4,000 0,949

LeftForearmGirth 23,711 3,800 1,056

RightWristGirth 15,632 2,900 0,791

LeftWristGirth 15,532 3,000 0,799

htWristDiamete

5,253 1,000 0,284

LeftWristDiamete

5,121 1,200 0,288

htU

erLe

Girth 54,226 12,400 3,269

LeftUpperLegGirth 53,847 12,200 3,363

RightLowerLegGirth 35,668 7,200 1,991

LeftLowerLegGirth 35,595 7,300 1,902

Tables 3 and 4 show basic descriptive parameters

(central – arithmetic mean, dispersive – range and

standard deviation) of the variables mesured digitally

by the BodyRecog PRO instrument.

Table 3: Descriptive parameters (digital measurement) –

male students.

Mean Range Std.Dev.

D-Bod

Mass 80,177 41,900 8,613

D-Bod

Hei

ht 184,570 36,107 7,646

D-AbdominalGirth 84,736 21,993 5,255

D-HipsGirth 101,436 22,833 4,766

D-ChestGirth 104,205 27,230 4,920

D-BreastGirth 97,753 23,763 4,577

D-Ri

htU

erArmG 33,221 10,673 2,536

D-LeftU

erArmG 32,673 10,397 2,617

D-Ri

htForearmG 30,525 15,037 2,737

D-LeftForearmG 31,118 33,343 6,276

D-RightWristGirth 19,789 11,057 2,207

D-LeftWristGirth 20,977 15,980 3,939

D-Ri

htU

erLe

G 54,709 31,893 5,496

D-LeftU

erLe

G 54,817 35,467 6,058

D-RightLowerLegG 37,770 19,913 2,982

D-LeftLowerLegG 37,226 20,037 3,043

Validity of a Structure Sensor-based Anthropometric Measurement: Performing a Pilot Study

247

We should emphasise here that the descriptive

parameters of the marked variables of the subsamples

of men and women (Tables 3 and 4) were computed

from the data saturated with the perceived and

recorded measurement errors. Unsuccessful scans

were reported primarily due to body movement -

examinee was not able to stand upright absolutely

still, low iPad battery or other types of software

issues.

The proportional contribution of the so

contaminated data to particular variables (no gender

differentiation) was the following: D-

RightForearmGirth (9.38%), D-LeftForearmGirth

(11.73%), D-RightWristGirth (13.14%), D-

LeftWristGirth (18.77%), D-RightUpperLegGirth

(3.28%), D-LeftUpperLegGirth (3.28%), D-

RightLowerLegGirth (10.79%), D-

LeftLowerLegGirth (10.79%). Although the research

was a pilot-project, the analysis results should be

observed with additional caution.

Table 4: Descriptive parameters (digital measurement) –

female students.

Mean Range Std.D.

D-BodyMass 61,953 28,300 7,267

D-BodyHeight 169,753 20,810 6,031

D-AbdominalGirth 81,307 18,793 4,951

D-Hi

sGirth 99,353 16,743 4,956

D-ChestGirth 90,936 19,363 4,660

D-BreastGirth 90,528 18,883 4,969

D-RightUpperArmG 30,266 8,520 1,905

D-LeftUpperArmG 30,229 8,353 2,016

D-Ri

htForearmG 26,233 10,493 2,576

D-LeftForearmGirth 26,148 7,310 1,757

D-RightWristGirth 17,490 7,800 2,257

D-LeftWristGirth 18,262 8,373 2,280

D-RightUpperLegG 51,769 26,203 5,687

D-LeftU

erLe

G 50,674 12,933 4,148

D-Ri

htLowerLe

G 39,829 16,203 4,359

D-LeftLowerLe

Girth 40,250 13,093 3,747

Deviation magnitudes

of the corresponding

variables (the ones with the matching measuring

points) in the group of the traditional and the digital

measurement are significantly different (Table 5).

The biggest deviation (in the form of average

increase in the results) was observed in the variable

delta-ChestGirth, followed by the variables of the

upper segments of the arms.

Only the circumferences of both the left and the

right thigh demonstrated a tendency of a significant

decrease in the results when compared with the

reference, traditional, measurement.

Prefix “delta” denotes a deviation magnitude.

It is interesting to notice that the first six variables,

whose measurement points are within the centrally

positioned reference framework of the instrument

(including the height of the instrument relative to the

measurement object), delta-BodyHeight, delta-

AbdominalGirth, delta-HipsGirth, delta-ChestGirth,

delta-BreastGirth, delta-RightUpperArmGirth, delta-

LeftUpperArmGirth, follow most proportinally

average deviations (with the increase in the results)

and adequate dispersion.

A higher dispersion of the

results was emphasised in every variable of the digital

circumference measurement positioned distally from

the body trunk and upper segments of the upper

extremities. The mentioned can also be followed via

the standard deviation magnitudes (Table 5).

Table 5: Deviation magnitudes.

Variable name Mean Ran

e Std.D.

delta

Bod

Mass -0,137 10,700 1,076

delta_BodyHeight 1,339 8,090 1,225

delta_AbdominalGirth 2,600 18,543 2,826

delta_HipsGirth 0,745 9,663 1,996

delta

ChestGirth 5,123 16,817 2,406

delta

BreastGirth 1,991 10,990 1,760

delta_RightUpperArmG. 2,064 4,910 1,142

delta_LeftUpperArmG. 2,019 7,120 1,479

delta_RightForearmG. 2,085 12,517 2,132

delta

LeftForearmGirth 2,891 28,920 4,892

delta

htWristGirth 2,044 9,607 2,097

delta

LeftWristGirth 3,270 14,837 3,332

delta_RightUpperLegG. -2,397 30,030 4,568

delta_LeftUpperLegG. -2,223 28,523 4,372

delta_RightLowerLegG. 0,540 19,173 3,649

delta

LeftLowerLe

G. 0,471 18,530 3,699

Table 6: Reliability measures.

Variable name

Crombach

Standardized

D-Bod

Hei

ht 0,989 0,990

D-AbdominalGirth 0,952 0,952

D-HipsGirth 0,897 0,898

D-ChestGirth 0,968 0,969

D-BreastGirth 0,987 0,988

D-Ri

htU

erArmG. 0,969 0,970

D-LeftU

erArmGirth 0,897 0,908

D-Ri

htForearmGirth 0,739 0,748

D-LeftForearmGirth 0,828 0,869

D-RightWristGirth 0,575 0,579

D-LeftWristGirth 0,672 0,696

D-Ri

htU

erLe

Girth 0,604 0,717

D-LeftU

erLe

Girth 0,621 0,721

D-RightLowerLegG. 0,454 0,523

D-LeftLowerLegGirth 0,539 0,574

K-BioS 2019 - Special Session on Kinesiology in Sport and Medicine: from Biomechanics to Sociodynamics

248

Reliability of the anthropometric measurement

using the digital instrument was expressed by the

method of internal consistency among the

measurement items; Cronbach’s and standardised

alpha were computed.

Reliability measures (Table 6) indicated an

acceptable reliability of the following digital girth

measures (variables): Body Height, AbdominalGirth,

HipsGirth, ChestGirth, BreastGirth,

RightUpperArmGirth, LeftUpperArmGirth. As

regards the other digital measures, a considerable

further work is needed.

The magnitudes of average correlation among the

items of digital measurement (which could also be

recognised as a homogeinity measure) expectedly

follow reliability decrements in case of the distal

measurement points.

Correlation coefficient magnitudes (Table 7)

indicate the correlation power of the corresponding

variables.

Table 7: Correlation magnitudes.

Classic

ital

Bod

Mass 0,996

D-Bod

Mass

Bod

Hei

ht 0,992

D-Bod

Hei

AbdominalGirth 0,863

D-AbdominalGirth

HipsGirth 0,912

D-HipsGirth

ChestGirth 0,949

D-ChestGirth

BreastGirth 0,951

D-BreastGirth

htU

erArmG 0,922

D-Ri

htU

erArmGirt

LeftU

erArmG. 0,875

D-LeftU

erArmGirth

RightForearmG. 0,763

D-RightForearmGirth

LeftForearmG. 0,578

D-LeftForearmGirth

htWristGirth 0,506

D-Ri

htWristGirth

LeftWristGirth 0,502

D-LeftWristGirth

htU

erLe

G 0,591

D-Ri

htU

erLe

Girth

LeftUpperLegG. 0,674

D-LeftUpperLegGirth

RightLowerLegG 0,244

D-RightLowerLegGirth

LeftLowerLegG. 0,212 D-LeftLowerLegGirth

The Pearson's correlation coefficient was used as

a measure of validity. Marked correlations (*) are

significant at p<0,05. An gradation of correlation

coefficients magnitudes (both in size and colour –

from cool colours to warm ones) clearly illustrates

association between the traditional measures and the

corresponding digital measures corroborating poorer

validity of distal measures in the comparison to the

central ones.

4 CONCLUSIONS

The analysed measurement instrument has not yet

met the targeted reliability level at all the measured

points (apart from, relatively, D-BodyHeight, D-

AbdominalGirt, D-ChestGirth, D-BreastGirth, D-

Hips Girth, D-RightUpperArmGirth and D-

LeftUpperArmGirth).

The obvious decrements in reliability of the

measures taken digitally distally from the body trunk

measures and upper segments of the upper extremities

indicate possible association with the technical

characteristics of the measuring instrument as well as

with the camera position management in relation to

the measurement object (the measures gathered at the

level of the central body trunk girth measures, with

no camera angle correction in relation to the measures

collected using the scanning angles corrections

towards the distal body segments while relatively

preserving the scanning height).

A needed additional partial analysis of varying

influences of measurement conditions and

techniques, as well as the analysis of their combined

influence on the measured results together with

additional software improvements will contribute to

the targeted measuring instrument’s utility.

The observed analytical limitations of the study

are closely related to the type of study conducted

(pilot study), therefore additional differences analysis

and standardized comparison methods will be made

after satisfactory hardware and software

modifications of the measuring instrument.

ACKNOWLEDGEMENTS

The research was conducted by the Joint Research

Group of Laboratory for Sports Medicine & Exercise

– Kinanthropometry and Biomechanics Laboratory of

the Institute of Kinesiology, Faculty of Kinesiology,

and companies Live Good d.o.o. Authors declare no

conflict of interest.

REFERENCES

Bragança, S., Carvalho, M., Xu, B.; Arezes, P., Ashdown,

S. A., 2014. Validation Study of a Kinect Based Body

Imaging (KBI) Device System Based on ISO

20685:2010. Conference. Lugano, 5th International

Conference on 3D Body Scanning Technologies.

Dell Statistica (data analysis software system), version 12.

software.dell.com.

Occipital, 2019. Technical Specification. Accessed May 15,

2019. https://support.structure.io/article/157-what-are-

the-structure-sensors-technical-specification

Rønnestad, B. R., Hansen, E. A., Raastad, T., 2010. Effect

of heavy strength training on thigh muscle cross-

sectional area, performance determi- nants, and

Validity of a Structure Sensor-based Anthropometric Measurement: Performing a Pilot Study

249

performance in well-trained cyclists. European Journal

of Applied Physiology, 108, 965–975.

doi:10.1007/s00421-009-1307-z

Ryder, J. R., Ball, S. D., 2012. Three-dimensional body

scanning as a novel technique for body composition

assessment: A preliminary investigation. J. Exerc.

Physiol. Online, 15(1):1-14.

Schranz, N. et al., 2010. ‘Three-dimensional

anthropometric analysis: Differences between elite

Australian rowers and the general population’, Journal

of Sports Sciences, 28(5), pp. 459–469. doi:

10.1080/02640411003663284.

Schranz, N. et al., 2012. ‘Is three-dimensional

anthropometric analysis as good as traditional

anthropometric analysis in predicting junior rowing

performance?’, Journal of Sports Sciences, 30(12), pp.

1241–1248. doi: 10.1080/02640414.2012.696204.

Simmons, K. P., Istook, C. L., 2003. Body measurement

techniques: A comparison of three-dimensional body

scanning and physical anthropometric methods for

apparel application. J. Fash. Mark. Manag., 7(3):306-

32.

Zhang, K.. Zheng, J., Gao, C., Thomas, D., Li, X.,

Heymsfield, S., 2014. Rapid-accurate anthropometric

body shape assessment with low- cost novel 3D

imaging system. F. A. S. E. B. J., 28(Suppl. 1):391.2, 2.

K-BioS 2019 - Special Session on Kinesiology in Sport and Medicine: from Biomechanics to Sociodynamics

250