Contents lists available at ScienceDirect

Acta Astronautica

Acta Astronautica 92 (2013) 79–88

0094-57

http://d

n Corr

Brook U

11794-5

E-m

journal homepage: www.elsevier.com/locate/actaastro

Prediction of trabecular bone qualitative properties usingscanning quantitative ultrasound

Yi-Xian Qin a,n, Wei Lin a, Erik Mittra a, Yi Xia a, Jiqi Cheng a, Stefan Judex a,Clint Rubin a, Ralph Muller b

a Department of Biomedical Engineering, Stony Brook University, NY 11794-5281, United Statesb Institute for Biomechanics, ETH Zurich, Zurich, Switzerland

a r t i c l e i n f o

Article history:

Received 30 November 2011

Received in revised form

2 July 2012

Accepted 24 August 2012Available online 5 October 2012

Keywords:

Quantitative ultrasound

Scanning confocal ultrasound

Ultrasound attenuation

Ultrasound velocity

Bone loss in space

Disuse osteoporosis

Bone adaptation

Bone mineral density

Speed of sound

Dual-energy X-ray absorptiometry

Bone quality

65/$ - see front matter & 2012 IAA. Publishe

x.doi.org/10.1016/j.actaastro.2012.08.032

espondence to: Department of Biomedical

niversity, Bioengineering Building, Rm 215

281, United States. Tel.: þ1 631 632 1481; fax

ail address: Yi-Xian.Qin@stonybrook.edu (Y.-X

a b s t r a c t

Microgravity induced bone loss represents a critical health problem in astronauts,

particularly occurred in weight-supporting skeleton, which leads to osteopenia and

increase of fracture risk. Lack of suitable evaluation modality makes it difficult for

monitoring skeletal status in long term space mission and increases potential risk of

complication. Such disuse osteopenia and osteoporosis compromise trabecular bone

density, and architectural and mechanical properties. While X-ray based imaging would

not be practical in space, quantitative ultrasound may provide advantages to char-

acterize bone density and strength through wave propagation in complex trabecular

structure. This study used a scanning confocal acoustic diagnostic and navigation

system (SCAN) to evaluate trabecular bone quality in 60 cubic trabecular samples

harvested from adult sheep. Ultrasound image based SCAN measurements in structural

and strength properties were validated by mCT and compressive mechanical testing.

This result indicated a moderately strong negative correlations observed between

broadband ultrasonic attenuation (BUA) and mCT-determined bone volume fraction

(BV/TV, R2¼0.53). Strong correlations were observed between ultrasound velocity (UV)

and bone’s mechanical strength and structural parameters, i.e., bulk Young’s modulus

(R2¼0.67) and BV/TV (R2

¼0.85). The predictions for bone density and mechanical

strength were significantly improved by using a linear combination of both BUA and

UV, yielding R2¼0.92 for BV/TV and R2

¼0.71 for bulk Young’s modulus. These results

imply that quantitative ultrasound can characterize trabecular structural and mechan-

ical properties through measurements of particular ultrasound parameters, and

potentially provide an excellent estimation for bone’s structural integrity.

& 2012 IAA. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Musculoskeletal deterioration and bone loss as well asassociated complications, i.e., disuse osteopenia and riskof fractures, are significant threats for astronauts duringlong-term space mission, e.g., in space station and the trip

d by Elsevier Ltd. All right

Engineering, Stony

, Stony Brook, NY

: þ1 631 632 8577.

. Qin).

to Mars [1]. Accumulated data from over 40 years of spaceexploration have demonstrated that space flight, particu-larly in long-term missions, has detrimental effects onbone and muscle. Results from short-term space mission(2–12 weeks) indicated that space flight with micrograv-ity alters calcium metabolism and bone mineral density(BMD) in several hundreds of men and women who haveflown in space. Osteopenia is a disease characterized bylong term loss of bone tissue, particularly in the weight-supporting skeleton [2]. On average, the magnitude andrate of the loss is staggering; astronauts lose bone mineral

s reserved.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–8880

in the lower appendicular skeleton at a rate approaching2% per month [3–7]. While osteopenia can affect the body,complications often occur predominantly at specific sitesof the skeleton with great load bearing demands. Thegreatest BMD losses in space have been observed in theskeleton of the lower body, i.e., in pelvic bones (�11.9971.22%) and in the femoral neck (�8.1771.24%), whilethere was no apparent decay found in the skull region[3–5]. Moreover, it is apparent that full recovery of bonemass may never occur [8–14], potentiating skeletal com-plications later in the astronaut’s life [11]. Similar resultswere found in the bedrest study. In a �61 head-down tilt7-day bed rest model for microgravity, it was observedthat there was a decreased bone formation rate in the iliaccrest [15]. Thus, assuming in a 2.5-year return-trip toMars, half of an astronaut’s bone density may vanish,severely jeopardizing their health and well-being. Follow-ing aging induced human osteoporosis pattern, it ispredicted that if a round trip to Mars would take 18–30months, significant bone loss would occur and impact theskeletal sustainability [16]. Moreover, the progressiveadaptation of the human biological system for short andlong term space flight still remain largely unknown, i.e.,current exercise countermeasure protocol cannot suffi-ciently prevent bone loss [17]. One of the reasons is theextreme difficulty in monitoring continuous adaptivedecay of bone loss during the space flight.

As human space exploration now plan and prepare toextend the mission to out orbital, such as human beingflight to Mars through extended manned vehicle with 18to 36 months of duration, one can imagine that the riskand the challenge in the musculoskeletal system will betremendous, and so little progress has been made inunderstanding the significance of the problem. There isalmost complete lack of on board measurements forassessing longitudinal bone loss and muscle atrophy, aswell as associated evaluation of countermeasure outcome.To elucidate microgravity affected skeleton disorders willlead to a better understanding of the barriers to long-termspace exploration and to assist in the development ofcountermeasures to assure safe and productive missions.To understand these effects, we need a better descriptionof human adaptation to space and with this informationcreate a prevention and countermeasure strategy throughnew technology. Developing new technologies will lead toa better understanding of the barriers to long-term spaceexploration and to assist in the development of counter-measures to assure safe and productive missions.

Bone mineral density (BMD) measurements are pre-dominantly used in the diagnosis of osteopenia andosteoporosis [18–21]. BMD measurement using dual-energy X-ray absorptiometry (DXA), commonly used inthe clinic, has several benefits, including relatively highprecision (�2%) and the capability of assessing severalindependent sites, such as the spine, hip and wrist.However, DXA is also limited, as the source and detectorare separated by the whole body, and the accuracy of themethod may be compromised, as it inherently includeslayers of soft tissue and is incapable of segregatingcortical from trabecular bone. DXA-determined BMD isan apparent two-dimensional (2D) index of the three-

dimensional (3D) structure rather than a true representa-tion of the 3D density. Thus, use of DXA to define thestructural quality of bone in vivo, whether in normal orosteoporotic subjects, remains largely unknown. Quanti-tative ultrasound (QUS) as a noninvasive modality tomeasure the peripheral skeleton has raised considerableinterest in bioengineering and clinical disciplines inrecent years [22]. New methods have emerged with thepotential to estimate not only the density of trabecularbone, but the strength and modulus of the bone tissue. QUSprovides an intriguing method for characterizing the mate-rial properties of bone in a manner which is noninvasive,non-ionized radiation, nondestructive, and also relativelyaccurate. Previous researches have demonstrated signifi-cance of QUS to quantify bone mass and structural stiffness[23–25]. Using several available clinical devices, studiesin vivo have shown the capability of QUS to discriminatepatients with osteoporotic fractures from age-matchedcontrols [26–30]. It has been demonstrated that QUS pre-dicts risk of future fracture generally as well as DXA [31,32].However, there are several notable limitations in the currentsetting of the ultrasound technologies, including the tissueboundary interaction, the nonlinear function of densityassociated with bone ultrasonic attenuation, a single indexcovering a broad range of tissues (including cortical andtrabecular regions), and the interpolation of the results.

Research efforts have been made using scanning mode ofthe ultrasound pulses in a 2-D or 3-D pattern [33,34–36].Such technology is intended to provide true images reflect-ing the structural and strength properties of bone atparticular skeletal sites in peripheral limbs and, in thefuture, has the potential to measure bones in deep tissues,like the great trochanter. This may further provide bothdensity and strength assessments in the regions of interestfor fracture risk [34,36]. QUS uses several parameters tocharacterize bone quantity and quality, i.e., ultrasonic wavepropagation velocity (UV), simple ultrasound attenuation(ATT), and broadband ultrasound attenuation (BUA), whichare closely related to acoustic transmission in a porousstructure that have been used to identify those individualsat risk of osteoporotic fracture as reliably as BMD [31,32,37].It has been shown that both BUA and UV are decreased inindividuals with risk factors for osteoporosis, i.e., primaryhyperparathyroidism [38–41], kidney disease [42], andglucocorticoid use [43,44].

To improve the specificity and resolution of imagebased quantitative ultrasonic assessment, a scanningconfocal acoustic navigation (SCAN) approach was devel-oped to identify the quality of trabecular bone. Theobjective of this study was to evaluate both bone struc-tural and strength properties using ultrasound scanning.The SCAN results were then validated using mCT-deter-mined morphologic properties and mechanical testing.

2. Methods

2.1. Sample preparation

A total of 60 trabecular bone cubes (1�1�1 cm3),were harvested from the distal femoral condyle of18 adult female sheeps (Warhill, intact ewes, 60–80 kg,

Fig. 1. Trabecular bone samples were carefully extracted from the sheep

distal femoral condyle. LG: Longitudinal direction in weight-bearing (451

with the long axis of bone); ML: Mediolateral direction; AP: Anteropos-

terior direction.

T R

xy

Fig. 2. Design of the scanning confocal acoustic diagnostic setup. Both

the transmitting (T) and receiving (R) transducers were co-focused at

the point of interest in the trabecular bone. The scan was performed so

that the middle plan in bone was scanned in 2D.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–88 81

6–8 years of age) that had previously undergone 2 years ofvibratory mechanical stimulation as part of an unrelatedresearch study[45] using a low-speed diamond blade saw(Microslice, Metals Research Limited, Cambridge, England)with continuous water cooling. 3 to 4 bone cubes wereharvested from each sheep. Prior to cutting, the femoralshaft was placed at a 451 angle to the blade (Fig. 1) so thatthe axes of the resultant samples corresponded to thephysiologic and anatomic directions, i.e., longitudinal (LG)(animal’s weight-bearing direction), anteroposterior (AP),and mediolateral (ML). Femoral rotation was standardizedfurther by positioning the bone so that the inferior surfacesof both condyles were equally in contact with the cuttingsurface. All harvestings were performed consistently usingstandard visual guidelines for all femoral cuts (e.g., theinferior-most transverse section was made just proximalto the intercondylar fossa) to ensure that the cubes wereharvested from the same relative location from each femur.

These bones were stored in a solution of half 70% ethanoland half normal saline at 4 1C with the marrow intact. Thisstorage method was chosen based upon the previous workof Ashman and colleagues, in which they harvested cancel-lous bone specimens for mechanical testing and ultrasoundstudies [23,46]. They performed specific experiments anddemonstrated that this solution preserved the elastic beha-vior of the ex vivo bone specimens for several months atroom temperature [47]. Using a nanomechanical testingmethod in our own studies, however, showed that after 12months of storage in this solution at 4 1C, an average 30%reduction in modulus and strength is observed [48]. For thisexperiment, the time from the first bone harvested to thelast bone mechanically tested was approximately fourmonths, with a mean time between harvesting and materi-als testing of only about one month per bone—well withinthe time frame analyzed by Ashman et al. The leachingof minerals over time that ultimately results in this loss ofbone strength, however, would not affect the geometry ofthe samples as visualized by mCT.

2.2. Quantitative ultrasound measurement

The specimen was submerged in a testing water tankfilled with degassed water. The bone cubes were measuredquantitatively using SCAN in three orthogonal directions.

The SCAN device consists of a computer-controlled 2Dscanner unit with an attached, focused transmitter andreceiver transducers (Panametrics V303, 1 MHz, 0.500 indiameter with target focus of 0.800; Olympus Panametrics,Billerica, MA). The specimen was placed in the middle of thetransducers and surrounded by sound-blocking materials.The transducers were coaxially aligned to each other inbone, so that the focal points for both the transmitting andreceiving ultrasound transducers converged at the point oftesting. Because we used a broadband transducer, even if itscenter frequency were 1 MHz, it would still have a broadfrequency response from 300 to 700 kHz, which is sufficientfor ultrasound measurement. The transmitter was driven bypulse signals, and the signals passing through bone werereceived by the receiver amplifier unit (Panametrics, Model5072PR) and digitized at 25 MHz using a high speeddigitizer (Gagescope, CS1250) embedded in a computer(Dell Dimension, Round Rock, TX). The control softwarewas written in the Cþþ language. The measurement proce-dure consisted of confocal scanning with the ultrasonicbeam through the central region (2D plane) of the samplewith a resolution of 0.5 mm pixel size (Fig. 2). A recording ofthe ultrasound wave was made over a 20�20 array(10�10 mm2 field of view). At first, the reference wavewas recorded without the bone sample in the ultrasoundpathway. Then the sample was inserted, and the test wavewas recorded. These waveforms were processed to calculatethe ATT (dB), the log-ratio of the energy of the referencewave to the test wave, as shown in Eq. (1).

ATT¼ 10log

Rs2

r ðtÞdtRs2

bðtÞdt

!ð1Þ

where sr(t) is the reference waveform and sb(t) is the bonewaveform. The BUA (dB/MHz) is defined as the slope of thefrequency-dependent attenuation within the 300 to700 kHz bandwidth, where the attenuation is linear [49].The UV was calculated using the time-of-flight method,which is based on the arrival time differences between thereference signal and the bone signal. The second zerocrossing point was used as the marker to identify the arrivalof the ultrasound wave. These ultrasonic values werefurther processed to generate images of ATT, BUA, and UV.A 14�14 grid (0.5 mm pixel size, 7�7 mm2 field of view)region of interest (ROI) was then determined from theimages of ATT, BUA, and UV to derive ultrasound

Fig. 3. Mechanical testing of bone cubic samples. The modulus of bone

was calculated by the linear slope of the strain–stress curve.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–8882

parameters. Averaged values of the above parameters werealso calculated.

Ultrasound attenuation in trabecular bone is usuallymeasured by the substitution method [50]. Accuracytesting of the BUA coefficient has been performed on aQUS normal heel phantom (QUS 304, ComputerizedImaging Reference Systems, Inc., Norfolk VA) which hassimilar acoustic properties to the human calcaneus. Thecalibrated BUA value of the QUS normal heel phantomwas 75.37 dB/MHz. Three repetitive measurements wereperformed on the phantom using the STM technique, andthe measured BUA values were recorded and compared tothe phantom calibration value. The coefficient of variation(CV¼standard deviation/mean) was also calculated.

2.3. mCT-determined bone microarchitecture and density

High-resolution mCT was used to evaluate the struc-ture and density of the bone samples [51–54]. A series ofstructural parameters of the trabecular samples, such asthe tissue volume (TV), bone volume (BV), bone volumefraction (BV/TV), mCT-based bone volumetric mineraldensity (vBMD) calculated by 3D-based bone volumefraction [55], mean trabecular thickness (Tb.Th), connec-tivity density (Conn.D), structural model index (SMI), anddegree of anisotropy (DA) [51] were determined from the3D images of the trabecular sample reconstructed with a34 mm resolution using built-in code (mCT-40, ScancoUSA., Southeastern, PA) [51,52,56]. The explanation ofthe various mCT indices and their respective calculationis found in several published sources [51,52,57,58] and sois not repeated here.

2.4. Tissue mechanical modulus

Contact force–displacement testing was used to deter-mine the mechanical properties of the trabecular bonesamples. Using a mechanical testing machine (MTS Sys-tems Corp., Eden Prairie, MN), the cubes were uniaxiallyloaded in compression using a displacement control(Fig. 3). To overcome slight deviations from surfaceparallelism, a smoothly curved nail head was placedabove the bone cube such that the force would bedistributed evenly to the bone in the loading direction[59] (Fig. 3). An upper limit of 300 N—determined byprior loading of non-experimental, but otherwise identi-cal, bone cubes—was established to prevent the plasticyielding of any specimens while the loading was achievedin bone’s elastic region. The loading rate was approxi-mately 1000 me per second for the samples. Both displa-cement and force were digitized and analyzed using MTSBasicTestware software. Prior to data collection, severalpreconditioning cycles at 1% strain were used to over-come edge effects from the harvesting process. Thispreconditioning consisted of at least five cycles and wasstopped once the preload stabilized at around 10 N orreached eight cycles (whichever came first). Subse-quently, three experimental compressions of 1% strainwere done, and the final result was taken to be theaverage of these three values.

Force–displacement was converted to an analogousstress–strain curve by dividing force by the cross-sectional area, and displacement by length (each cubewas measured independently to reduce the geometricalerror inherent in harvesting).

The material properties studied included elastic moduliin three orthogonal directions and bulk modulus, which wasthe averaged value of the elastic moduli from the threeorthogonal directions. After scanning in the mCT, the boneswere tested until failure in the LG direction. The materialproperties studied include modulus, yield strength (calcu-lated using the 0.2% strain offset method), and ultimatestrength.

2.5. Correlations and statistical analyses

Interrelationships between QUS parameters and mCT-determined structural values, and between QUS para-meters and mechanical properties were evaluatedthrough multiple correlations using Analyze-it (version1.67; Leeds, UK) and SPSS (version 16; Chicago, IL).Statistical correlation was performed between the ultra-sound parameter and the trabecular structural para-meters and mechanical parameters. When ultrasoundparameters were correlated to Young’s modulus of thetrabecular bone in three orthogonal directions, the ultra-sound velocity and BUA in the corresponding directionswere used. Otherwise, if the bone parameter was notdirectionally dependent (e.g., density), the averaged ultra-sound velocity and BUA from the three orthogonal direc-tions were used. Finally, a combined linear regression ofBUA and UV was used to interpret the complex structureof trabecular bone and its interactive influence on thederived ultrasound signals. The data were analyzed usingPearson product-moment correlation coefficients, and thesignificance level was set at po0.05.

3. Results

The microarchitecture and mechanical properties andthe ultrasound measurement results of trabecular bonecubes varied among the samples (Tables 1–3) (Fig. 4).Table 1 shows the mechanical properties from the threeanatomic loading directions, AP, LG, and ML. The mechan-ical stiffness of sheep trabecular bone demonstrateddirectional variation from 512796 MPa in LG to442797 MPa in AP and to 3467100 MPa in ML, yieldinga bulk modulus of 433793 MPa and a yield strength of

Table 1Mechanical testing data for bone cubes (MPa).

E-AP E-LG E-ML Bulk E Yield Strength

mean7SD 441.52796.55 511.67796.16 345.767100.12 432.98792.52 16.8576.78

CV(%) 21.9 18.8 29.0 21.4 40.2

Median 416.86 516.84 337.54 423.14 16.53

Minimum 246.75 311.48 142.90 253.61 7.48

Maximum 691.66 716.71 548.04 641.77 35.38

Table 2Micro-CT data for bone cubes.

BV/TV SMI Conn.D (1/mm3) Tr.Th. (mm) DA

mean7SD 0.2970.07 0.5670.40 4.8671.01 0.1970.05 1.8870.20

CV(%) 24.1 70 20.8 23.8 10.8

Median 0.28 0.53 4.79 0.18 1.87

Minimum 0.17 0.01 2.81 0.15 1.42

Maximum 0.44 1.59 6.78 0.45 2.37

Table 3Ultrasound testing for bone cubes, mean7SD.

ATT (dB/cm) BUA (dB/MHz/cm) UV (m/s)

AP 25.1973.67 111.91735.55 2122.167133.44

LG 29.4273.52 120.61735.97 2437.247197.14

ML 20.8872.85 74.66713.36 1883.137109.63

Average 25.1673.04 102.39723.07 2147.517133.75

Mean CV(%) 12.1 22.5 6.2

Bulk median 24.46 104.51 2147.42

Bulk min 20.10 58.10 1805.75

Bulk max 33.26 151.13 2418.84

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–88 83

1776.8 MPa. Among all the directions, the elastic mod-ulus in the LG direction was greater than the ones in theML and AP directions.

The microstructural properties of trabecular bonesamples, in which bone volume fraction (BV/TV) variedfrom 17% (minimum) to 44% (maximum). Similar varia-tions were found in SMI, Conn.D, Tr.Th, and DA (Table 2).

Ultrasound measurements (ATT, BUA, and UV) showedthe directional variations from anatomic positions, AP, LG,and ML (Table 3). BUA and ATT values in the LG directionwere greater than the ultrasound values in the ML and APdirections. ATT in the LG direction was 16.8% and 40.9%greater than in the AP and the ML directions, respectively.Similarly, BUA in the LG direction was 7.8% and 61.5%greater than in the AP and the ML directions, respectively.UV in the longitudinal direction (LG) was greater than inthe AP (14.8%) and ML (29.4%) directions.

Table 4 shows the overall correlation coefficients (R).A strong correlation was found between BUA and BV/TV(R2¼0.53, po0.001, Fig. 5). A higher correlation was

found between average ultrasound velocity and theBV/TV (R2

¼0.86, po0.001, Fig. 6). Strong correlationswere found between UV and bone strength and structuralparameters such as bulk Young’s modulus (R2

¼0.67),BV/TV (R2

¼0.85), and Tb.Th (R2¼0.48). Although weak

compared to the UV results, correlations between BUAand mCT-determined structural parameters, such as BV/TV(R¼�0.73 or R2

¼0.53), and Tb.Th (R¼�0.34 or

R2¼0.12), as well as tissue bulk modulus (R¼�0.31 or

R2¼0.12), were still high. Also, the variability of the

correlation coefficients between BUA and the mechanicalmodulus was greater than the variability between ultra-sound velocity and the mechanical modulus. The correla-tions were improved by using a new parameter thatcombined BUA and UV in a linear regression analysis,yielding values of R2

¼0.92 for BV/TV, R2¼0.45 for trabe-

cular thickness, and R2¼0.71 for bulk modulus. The linear

combination of ultrasound velocity and BUA predicted70% of the variation in the bulk Young’s modulus(R2¼0.70, Fig. 7) and 87% of the variation in the BV/TV

(R2¼0.87, Fig. 8). Combined BUA and UV indexes

improved the correlation coefficients consistently above0.75 with regard to the elastic Young’s modulus, bulkmodulus, yield, and ultimate strength (Table 4). Strongcorrelations were also found between this combined BUAand the UV from a single direction, i.e., ML, and thestructural and strength parameters.

Strong correlations existed between the combinedBUA and UV in either the ML or AP direction, as well asthe overall structural and mechanical properties (Table 4,columns 5 and 6). Significant correlations were foundbetween the combined BUA and UV (in ML and APdirections) and the bulk modulus (R2

¼0.58 andR2¼0.56), yielding strength (R2

¼0.76 and R2¼0.79),

BV/TV (R2¼0.86), and SMI (R2

¼0.83 and R2¼0.85),

respectively. This indicates that QUS measured in thenon-longitudinal directions can predict overall mechan-ical and structural properties of trabecular bone.

4. Discussion

QUS parameters (e.g., BUA and UV) measured by theSCAN have shown strong correlation with the structural andmechanical properties of the trabecular bone samples. Theresults have demonstrated that the correlations betweenQUS parameters and mCT-determined volumetric bonedensity (BV/TV) reaches strong agreement as high as 70%to 90% [60–62]. Use of mCT as a structural assessmentmodality does not compromise the potential of QUS in bone

Fig. 4. mCT-measured trabecular bone structure, 34 mm resolution. Left is the normal bone; right is the sample with low bone density.

Table 4Relative correlation coefficients (R values) for QUS, mCT and mechanical testing. SCAN determined ultrasound parameters are able to predict structural

and strength properties of trabecular bone. Combined BUA and UV indexes improve the correlation coefficients consistently above 0.75 with regard to the

elastic modulus, bulk modulus, yield or ultimate strength. Strong correlations were also found between combined BUA and UV from single direction, i.e.,

ML, and structural and strength parameters, suggesting QUS signals extracted from particular direction (e.g., under in vivo condition) are able to predict

overall trabecular bone density and stiffness.

ATT UV BUA Combo BUA &

UV in ML

Combo BUA &

UV in AP

Combo BUA & UV

AP Modulus �0.71 0.79 �0.31 0.74 0.74 0.80

po0.001 po0.001 p¼0.033 po0.001 po0.001 po0.001

LG Modulus �0.75 0.79 �0.221 0.63 0.67 0.81

po0.001 po0.001 p¼0.131 po0.001 po0.001 po0.001

ML Modulus �0.673 0.89 �0.483 0.79 0.74 0.78

po0.001 po0.001 p¼0.001 po0.001 po0.001 po0.001

Bulk Modulus �0.75 0.82 �0.358 0.76 0.75 0.84

po0.001 po0.001 p¼0.012 po0.001 po0.001 po0.001

Yield Strength �0.72 0.90 �0.85 0.87 0.89 0.93

po0.001 po0.001 po0.001 po0.001 po0.001 po0.001

Ulti. Strength �0.75 0.90 �0.85 0.87 0.88 0.94

po0.001 po0.001 po0.001 po0.001 po0.001 po0.001

BV/TV �0.37 0.93 �0.73 0.93 0.93 0.93

p¼0.018 po0.001 po0.001 po0.001 po0.001 po0.001

BMD 0.74 0.85 �0.75 0.86 0.85 0.87

SMI 0.3 0.9 0.66 0.91 0.92 0.93

p¼0.034 po0.001 po0.001 po0.001 po0.001 po0.001

Conn.D. �0.12 �0.33 0.07 0.32 0.39 0.42

p¼0.5 p¼0.006 p¼0.5 p¼0.025 p¼0.006 p¼0.003

Tr.Th. �0.17 0.69 �0.34 0.83 0.64 0.67

p¼0.368 po0.001 p¼0.01 po0.001 po0.001 po0.001

DA �0.278 0.5 �0.32 0.34 0.44 0.50

p¼0.056 po0.001 p¼0.26 p¼0.17 p¼0.002 po0.001

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–8884

quality assessment, where ultrasound technology—whichnot only measures density but also bone strength—has beenproven to be safe, noninvasive, non-ionizing, portable, andrelatively inexpensive. This would provide a potential mod-ality for onboard bone quality assessment in the short- andlong-duration space missions, and monitoring progressivebone remodeling as well as evaluating effects of variouscountermeasure therapies.

QUS has demonstrated directional sensitive to trabe-cular orientation [63,64], which can provide extra infor-mation in the architecture of trabecular anisotropy and onhow ultrasound parameters are associated with the DA.Strong correlations were found between combined BUAand the UV from a single direction, i.e., ML, and thestructural and strength parameters, suggesting that QUSsignals extracted from a particular direction under in vivo

Fig. 5. The correlation of average BUA with the BV/TV from mCT analysis.

R2 is 0.53 (po0.001).

Fig. 6. The correlation of averaged UV with the bulk modulus. R2 is 0.63

(po0.001).

Fig. 7. Prediction of bulk modulus by the linear combination of ultra-

sound velocity and BUA also showed high correlation (R2¼0.7,

po0.001). E(predict)¼0.625UVþ0.585BUAþ968.

Fig. 8. Prediction of BV/TV by the linear combination of ultrasound

velocity and BUA showed high correlation (R2¼0.87, po0.001).

BV/TV(predict)¼3.89�10�4UV�9�10�4BUA�0.456.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–88 85

conditions would be able to predict principal trabecularorientation and therefore to estimate overall stiffness oftrabecular bone. In our recently published data, QUS iscapable to predict trabecular bone principal structuraldirection close enough with mCT determined MIL tensoras small as 51 [64]. In addition, QUS BUA and UV canpredict density variations, in which combination of BUAand UV can predict bone volume fraction as high as 86%.QUS results are closely correlated to other architectureparameters of trabecular bone as well, e.g., Tb.Th and SMI.These results suggest that QUS parameters are influencedby both bone density and architecture. Among measured

QUS parameters, significant correlation is observed betweenUV and mechanical strength, suggesting UV may be a strongcandidate parameter closely related to bone stiffness. Inorder to predict bone’s structure and mechanical propertiesusing QUS, the contributions of multiple ultrasound para-meters should be considered. Indeed, if a functional rela-tionship between multiple QUS parameters and measuredbone density and stiffness could be generated by performinga large number of sample measurements across species,QUS can theoretically predict both the structural andstrength properties of bone.

In this study, averaged BUA was inversely correlatedwith BV/TV (Fig. 5). In previous human QUS measurementin osteoporotic bone, the relationship between BUA andbone mineral density has shown positive correlation[62,65,66]. Usually, in human trabecular bone, the poros-ity varies in the range of 75% to 95% (equivalent to 5% to25% of bone volume fraction) depending on the status ofbone quality [28,52]. However, normal human bone andanimal trabeculae, e.g., sheep bone, can have a relativelyhigher bone volume fraction than aged human bone,which results in a trabecular bone volume fraction inthe range of 25% to 50%. In low-density sheep trabecularbone (more porous, like the human calcaneus), BUA seemsto mainly reflect the effect of absorption due to a relativelysmaller interaction of sound waves with trabecular archi-tecture than with dense trabecular bone [67]. On the otherhand, in dense trabecular bone, the effect of scattering ofultrasound waves on BUA may be highly dominant as aresult of more interaction of the ultrasound within trabe-cular porosity. This relation was demonstrated by an experi-mental setup for the interrelationship between BUA and awide range of bone mineral densities [67], in which bothbone mass and structure governed the BUA in a complicatednonlinear relation. This complex relation can also beexplained from our previous work using a modified strati-fied model [49], in which BUA demonstrated a parabolicnonlinear shape against the porosity [68–70]. These datasuggest that BUA is a strong indicator of trabecular bonemass and architecture. However, interpreting the BUA datawith respect to bone structure parameters should be donewith extra caution, as BUA depends strongly on the bonevolume fraction and porosity values.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–8886

The image-based QUS measurement of bone samples iscapable to improve the correlations with bone structure andmechanical properties, thereby revealing bone quality infor-mation in the region of interest. In this study, the extractedtrabecular bone samples excluded cortical bone. Althoughthe potential influence of the cortical shell and of irregularbone surfaces were not discussed in this study, these factorscan be addressed, and their relative influences in measuredultrasound parameters can be analyzed in a future study.The influences of cortical shell and irregular bone surfaceshave been evaluated in separate studies [62,71], whichdemonstrated that this new scanning ultrasound technologywas capable of identifying regions of interest and theirfeatures of inhomogeneity. The variations induced by anirregular shape and cortical shell can be minimized. Thus, itis possible to make noninvasive measurements of largebone sites, such as the calcaneus, for clinical assessment.Furthermore, the SCAN system can identify the inhomo-geneity in bone and predict the sites at risk of weakness bymeans of its scanning feature.

Although QUS is capable of predicting the density andmechanical properties of bone, QUS does not directly mea-sure such bone properties such as density and modulus [24].Thus, a well-established database for the interrelationshipbetween QUS and bone structural and strength propertieswould provide insight into the noninvasive diagnosis of bonequality using such ultrasound methods. Use of both mCT andmechanical testing on the trabecular samples has helped togenerate such functional relation and makes the noninvasiveQUS measurement for bone quality possible as a diagnostictool. The ultrasound values correlated best with overallparameters such as bone volume fraction and SMI, whichare the best indicators of global quality of the bone, ratherthan with such specific parameters as Conn.D, density, andTb.Th. Ultrasound values correlated better with yield andultimate strength, the best indicators of true fracture risk,than with elastic modulus, a simple measure of stiffness.

Moreover, the measurements of QUS and mCT in thisstudy were conducted only in the sheep bone withparticular porosity and density. Although relativelyaltered bone structure and strength parameters in thesesamples were identified, it would not fully represent thetrue status of normal and osteoporotic trabecular condi-tions in human. Future research is required to determinethe correlation between measured ultrasound images andtrue bone properties, such as osteopenia and osteoporosiswith significantly low density and high porosities oftrabecular structure, using various and more realisticsamples and/or phantoms. These may point to a nicedesign in future works and manuscripts.

5. Conclusion

Quantitative ultrasound has demonstrated promisingpotentials in the nondestructive assessment of bone struc-tural and strength parameters. With image based QUS, it ispotential for in vivo bone quality assessment, as the initialsites of bone loss in osteopenia and osteoporosis occurprimarily in the trabecular region. In addition to predictingthe density and structure properties of bone, the SCAN QUSdemonstrates an encouragingly high correlation between

bone modulus and ultrasound parameters, suggesting thatQUS may provide a strong estimation for the noninvasiveevaluation of strength, an important integrity parameter forpredicting the risk of fracture. QUS measurement of thetrabecular bone is directionally sensitive, consistent withthe structural and mechanical properties of bone. Theresults suggest that QUS scanning can characterize bonequality in the region of interest identified by QUS generatedimaging. Although, ultrasound is not a direct measurementfor either trabecular structure or mechanical modulus, withan accumulated database of normal and diseased boneassessment, QUS has the potential to identify osteoporosisand fracture risk and to monitor skeletal regenerativeadaptation as a potential onboard modality for long-termspace mission.

Acknowledgments

This work is kindly supported by the National SpaceBiomedical Research Institute (SMST01603, Qin) throughNASA Cooperative Agreement NCC 9-58, NIH (AR49286and AR52379, Qin), and NYSTAR (Qin).

References

[1] T. Lang, A. LeBlanc, H. Evans, Y. Lu, H. Genant, A. Yu, Cortical andtrabecular bone mineral loss from the spine and hip in long-duration spaceflight, J. Bone Miner. Res. 19 (2004) 1006–1012.

[2] B.L. Riggs, L.J. Melton III, The worldwide problem of osteoporosis:insights afforded by epidemiology, Bone 17 (1995) 505S–511S.

[3] A. LeBlanc, V. Schneider, L. Shackelford, Bone mineral and leantissue loss after long duration spaceflight, Trans. Am. Soc. BoneMiner. Res. 11S (1996) 567.

[4] A. LeBlanc, L. Shackelford, A. Feiveson, V. Oganov, Bone loss inspace: shuttle/mir experience and bed-rest counter measure pro-gram, in: Proceedings of the 1st Biennial Space Biomedical Inves-tigators’ Workshop, 1999, p. 17.

[5] C. Ruff, T. Beck, D. Newman, M. Oden, G. Shaffner, A. LeBlanc,L. Shackelford, N. Rianon, Skeletal consequences of reduced gravityenvironments, in: Proceedings of the 1st Biennial Space BiomedicalInvestigators’ Workshop, 1999, pp. 86–87.

[6] A. LeBlanc, V. Schneider, L. Shackelford, S. West, V. Oganov,A. Bakulin, L. Voronin, Bone mineral and lean tissue loss after longduration space flight, J. Musculoskelet. Neuronal. Interact. 1 (2000)157–160.

[7] A.D. LeBlanc, E.R. Spector, H.J. Evans, J.D. Sibonga, Skeletalresponses to space flight and the bed rest analog: a review,J. Musculoskelet. Neuronal. Interact. 7 (2007) 33–47.

[8] A. Goode, Musculoskeletal change during spaceflight: a new view ofan old problem, Br. J. Sports Med. 33 (1999) 154.

[9] A.W. Goode, P.C. Rambaut, The skeleton in space, Nature 317 (1985)204–205.

[10] A. LeBlanc, V. Schneider, H. Evans, D. Engelbretson, J. Krebs, Bonemineral loss and recovery after 17 weeks of bed rest, J. Bone Miner.Res. 5 (8) (1990) 843–850.

[11] A. LeBlanc, V. Schneider, Can the adult skeleton recover lost bone?Exp. Gerontol. 26 (1991) 189–201.

[12] P. Rambaut, A. Goode, Skeletal changes during space flight, Lancet 2(1985) 1050–1052.

[13] L. Shackleford, A. LeBlanc, A. Feiveson, V. Oganov, Bone loss inspace: Shuttle/Mir experience and bed rest countermeasure pro-gram, in: Proceedings of the 1st Biennial Space Biomedical Inves-tigators’ Workshop, vol. 1, 1999, pp. 86–87.

[14] F. Tilton, J. Degioanni, V. Schneider, Long term follow up of SkyLabbone demineralization, Aviat. Space Environ. Med. 51 (1980)1209–1213.

[15] S.B. Arnaud, D.J. Sherrard, N. Maloney, R.T. Whalen, P. Fung, Effects of1-week head-down tilt bed rest on bone-formation and the calciumendocrine system, Aviat. Space Environ. Med. 63 (1992) 14–20.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–88 87

[16] Y.X. Qin, Challenges to the musculoskeleton during a journey toMars: assessment and counter measures, J. Cosmol. 12 (2010)3778–3780.

[17] A.I. Grigoriev, V.S. Oganov, A.V. Bakulin, V.V. Poliakov, L.I. Voronin,V.V. Morgun, V.S. Shnaider, L.V. Murashko, V.E. Novikov, A. LeBlanc,L. Shackelford, Clinical and psychological evaluation of bonechanges among astronauts after long term space flights, Aviakosm.Ekolog. Med. 32 (1) (1998) 21–25. (in Russian).

[18] H.K. Genant, Current state of bone densitometry for osteoporosis,Radiographics 18 (1998) 913–918.

[19] J.A. Kanis, F. Borgstrom, N. Zethraeus, O. Johnell, A. Oden,B. Jonsson, Intervention thresholds for osteoporosis in the UK, Bone36 (2005) 22–32.

[20] L.J. Melton III, E.S. Orwoll, R.D. Wasnich, Does bone density predictfractures comparably in men and women? Osteoporos. Int. 12(2001) 707–709.

[21] L.J. Melton III, J.A. Kanis, O. Johnell, Potential impact of osteoporosistreatment on hip fracture trends, J. Bone Miner. Res. 20 (2005)895–897.

[22] P. Laugier, Quantitative ultrasound of bone: looking ahead, JointBone Spine 73 (2006) 125–128.

[23] R.B. Ashman, J.Y. Rho, Elastic modulus of trabecular bone material,J. Biomech. 21 (1988) 177–181.

[24] M. Harada, K. Tanaka, T. Katayama, K. Mizuno, H. Soumiya,M. Matsukawa, Relationship between mechanical properties andacoustic parameters obtained from fast and slow waves for can-cellous bone, J. Acoust Soc. Am. 123 (2008) 3633.

[25] P.H. Nicholson, R. Alkalay, Quantitative ultrasound predicts bonemineral density and failure load in human lumbar vertebrae, Clin.Biomech. (Bristol., Avon.) 22 (2007) 623–629.

[26] S. Cheng, F. Tylavsky, L. Carbone, Utility of ultrasound to assess riskof fracture, J. Am. Geriatr. Soc. 45 (1997) 1382–1394.

[27] E.W. Gregg, A.M. Kriska, L.M. Salamone, M.M. Roberts, S.J. Anderson,R.E. Ferrell, L.H. Kuller, J.A. Cauley, The epidemiology of quantitativeultrasound: a review of the relationships with bone mass, osteoporosisand fracture risk, Osteoporos. Int. 7 (1997) 89–99.

[28] C.F. Njeh, I. Saeed, M. Grigorian, D.L. Kendler, B. Fan, J. Shepherd,M. McClung, W.M. Drake, H.K. Genant, Assessment of bone statususing speed of sound at multiple anatomical sites, Ultrasound Med.Biol. 27 (2001) 1337–1345.

[29] P.H. Nicholson, R. Strelitzki, On the prediction of Young’s modulusin calcaneal cancellous bone by ultrasonic bulk and bar velocitymeasurements, Clin. Rheumatol. 18 (1999) 10–16.

[30] J. Damilakis, G. Papadokostakis, H. Vrahoriti, I. Tsagaraki,K. Perisinakis, A. Hadjipavlou, N. Gourtsoyiannis, Ultrasoundvelocity through the cortex of phalanges, radius, and tibia innormal and osteoporotic postmenopausal women using a newmultisite quantitative ultrasound device, Invest. Radiol. 38 (2003)207–211.

[31] D.C. Bauer, C.C. Gluer, J.A. Cauley, T.M. Vogt, K.E. Ensrud,H.K. Genant, D.M. Black, Broadband ultrasound attenuation pre-dicts fractures strongly and independently of densitometry in olderwomen. A prospective study. Study of Osteoporotic FracturesResearch Group, Arch. Intern. Med. 157 (1997) 629–634.

[32] D. Hans, C.F. Njeh, H.K. Genant, P.J. Meunier, Quantitative ultra-sound in bone status assessment, Rev. Rhum. Engl. Ed. 65 (1998)489–498.

[33] P. Laugier, V. Novikov, B. Elmann-Larsen, G. Berger, Quantitativeultrasound imaging of the calcaneus: precision and variationsduring a 120-Day bed rest, Calcif. Tissue Int. 66 (2000) 16–21.

[34] Y.-X. Qin, W. Lin, C. Rubin, Interdependent relationship betweentrabecular bone quality and ultrasound attenuation and velocityusing a scanning confocol acoustic diagnostic system, J. BoneMiner. Res. 16 (2001) S470.

[35] Y.X. Qin, Y. Xia, W. Lin, J. Cheng, J. Muir, C. Rubin, Longitudinalassessment of human bone quality using scanning confocal quan-titative ultrasound, J. Acoust Soc. Am. 123 (2008) 3638.

[36] Y. Xia, W. Lin, Y. Qin, The influence of cortical end-plate onbroadband ultrasound attenuation measurements at the humancalcaneus using scanning confocal ultrasound, J. Acoustic Soc. Am.118 (2005) 1801–1807.

[37] M.L. Frost, G.M. Blake, I. Fogelman, Quantitative ultrasound andbone mineral density are equally strongly associated with riskfactors for osteoporosis, J. Bone Miner. Res. 16 (2001) 406–416.

[38] S. Gonnelli, A. Montagnani, C. Cepollaro, R. Monaco, L. Gennari, B. Rossi,S. Pacini, C. Gennari, Quantitative ultrasound and bone mineral densityin patients with primary hyperparathyroidism before and after surgicaltreatment, Osteoporos. Int. 11 (2000) 255–260.

[39] B.M. Ingle, W.E. Thomas, R. Eastell, Differential effects of primaryhyperparathyroidism on ultrasound properties of bone, Osteo-poros. Int. 13 (2002) 572–578.

[40] S. Minisola, R. Rosso, A. Scarda, M.T. Pacitti, E. Romagnoli, G. Mazzuoli,Quantitative ultrasound assessment of bone in patients with primaryhyperparathyroidism, Calcif. Tissue Int. 56 (1995) 526–528.

[41] S. Minisola, M.T. Pacitti, E. Ombricolo, G. Costa, A. Scarda,E. Palombo, R. Rosso, Bone turnover and its relationship with bonemineral density in pre- and postmenopausal women with orwithout fractures, Maturitas 29 (1998) 265–270.

[42] A. Wittich, E. Vega, C. Casco, A. Marini, C. Forlano, F. Segovia,M. Nadal, C. Mautalen, Ultrasound velocity of the tibia in patientson haemodialysis, J. Clin Densitom. 1 (1998) 157–163.

[43] F. Blanckaert, B. Cortet, P. Coquerelle, R.M. Flipo, B. Duquesnoy,X. Marchandise, B. Delcambre, Contribution of calcaneal ultrasonicassessment to the evaluation of postmenopausal and glucocorticoid-induced osteoporosis, Rev. Rhum. Engl. Ed. 64 (1997) 305–313.

[44] B. Cortet, R.M. Flipo, F. Blanckaert, B. Duquesnoy, X. Marchandise,B. Delcambre, Evaluation of bone mineral density in patients withrheumatoid arthritis. Influence of disease activity and glucocorti-coid therapy, Rev. Rhum. Engl. Ed. 64 (1997) 451–458.

[45] C. Rubin, A.S. Turner, R. Muller, E. Mittra, K. McLeod, W. Lin,Y.X. Qin, Quantity and quality of trabecular bone in the femur areenhanced by a strongly anabolic, noninvasive mechanical inter-vention, J. Bone Miner. Res. 17 (2002) 349–357.

[46] R.B. Ashman, P.P. Antich, J. Gonzales, J.A. Anderson, J.Y. Rho,A comparison of reflection and transmission ultrasonic techniquesfor measurement of cancellous bone elasticity, J. Biomech. 27(1994) 1195–1199.

[47] R.B. Ashman, S.C. Cowin, W.C. Van Buskirk, J.C. Rice, A continuouswave technique for the measurement of the elastic properties ofcortical bone, J. Biomech. 17 (1984) 349–361.

[48] E. Mittra, S. Akella, Y.X. Qin, The effects of embedding material,loading rate and magnitude, and penetration depth in nanoinden-tation of trabecular bone, J. Biomed. Mater. Res. A 79 (2006) 86–93.

[49] W. Lin, Y.X. Qin, C. Rubin, Ultrasonic wave propagation in trabe-cular bone predicted by the stratified model, Ann. Biomed. Eng. 29(2001) 781–790.

[50] C.M. Langton, C.F. Njeh, Acoustic and ultrasonic tissue char-acterization—assessment of osteoporosis, Proc. Inst. Mech. Eng. H213 (1999) 261–269.

[51] T. Hildebrand, P. Ruegsegger, Quantification of bone microarchi-tecture with the structure model index, Comput. Methods Biomech.Biomed. Eng. 1 (1997) 15–23.

[52] T. Hildebrand, A. Laib, R. Muller, J. Dequeker, P. Ruegsegger, Directthree-dimensional morphometric analysis of human cancellousbone: microstructural data from spine, femur, iliac crest, andcalcaneus, J. Bone Miner. Res. 14 (1999) 1167–1174.

[53] R. Muller, G.H. van Lenthe, Trabecular bone failure at the micro-structural level, Curr. Osteoporos. Rep. 4 (2006) 80–86.

[54] A. Odgaard, Three-dimensional methods for quantification of can-cellous bone architecture, Bone 20 (1997) 315–328.

[55] R. Muller, G.H. van Lenthe, Trabecular bone failure at the micro-structural level, Curr. Osteoporos. Rep. 4 (2006) 80–86.

[56] P. Ruegsegger, B. Koller, R. Muller, A microtomographic system forthe nondestructive evaluation of bone architecture, Calcif. TissueInt. 58 (1996) 24–29.

[57] P. Ruegsegger, B. Koller, R. Muller, A microtomographic system forthe nondestructive evaluation of bone architecture, Calcif. TissueInt. 58 (1996) 24–29.

[58] R.J. Fajardo, R. Muller, Three-dimensional analysis of nonhumanprimate trabecular architecture using micro-computed tomography,Am. J. Phys. Anthropol. 115 (2001) 327–336.

[59] E. Mittra, C. Rubin, Y.X. Qin, Interrelationship of trabecular mechanicaland microstructural properties in sheep trabecular bone, J. Biomech.38 (2005) 1229–1237.

[60] G. Haiat, F. Padilla, F. Peyrin, P. Laugier, Variation of ultrasonicparameters with microstructure and material properties of trabecularbone: a 3D model simulation, J. Bone Miner. Res. 22 (2007) 665–674.

[61] M.A. Hakulinen, J.S. Day, J. Toyras, H. Weinans, J.S. Jurvelin, Ultra-sonic characterization of human trabecular bone microstructure,Phys. Med. Biol. 51 (2006) 1633–1648.

[62] Y. Xia, W. Lin, Y.X. Qin, Bone surface topology mapping and its rolein trabecular bone quality assessment using scanning confocalultrasound, Osteoporos. Int. 18 (2007) 905–913.

[63] G. Luo, J.J. Kaufman, A. Chiabrera, B. Bianco, J.H. Kinney, D. Haupt,J.T. Ryaby, R.S. Siffert, Computational methods for ultrasonic boneassessment, Ultrasound Med. Biol. 25 (1999) 823–830.

Y.-X. Qin et al. / Acta Astronautica 92 (2013) 79–8888

[64] L. Lin, J. Cheng, W. Lin, Y.-X. Qin, Prediction of trabecular boneprincipal structural orientation using quantitative ultrasound scan-ning, J. Biomech. 45 (2012) 1790–1795.

[65] M. Bolanowski, D. Jedrzejuk, A. Milewicz, A. Arkowska, Quantitativeultrasound of the heel and some parameters of bone turnover inpatients with acromegaly, Osteoporos. Int. 13 (2002) 303–308.

[66] V.W. Hung, L. Qin, S.K. Au, W.Y. Choy, K.S. Leung, P.C. Leung,J.C. Cheng, Correlations of calcaneal QUS with pQCT measurementsat distal tibia and non-weight-bearing distal radius, J. Bone Miner.Metab. 22 (2004) 486–490.

[67] S. Han, J. Rho, J. Medige, I. Ziv, Ultrasound velocity and broadbandattenuation over a wide range of bone mineral density, OsteoporosInt. 6 (1996) 291–296.

[68] L. Serpe, J.Y. Rho, The nonlinear transition period of broadbandultrasound attenuation as bone density varies, J. Biomech. 29

(1996) 963–966.

[69] L.J. Serpe, J.Y. Rho, Broadband ultrasound attenuation value depen-dence on bone width in vitro, Phys. Med. Biol. 41 (1996) 197–202.

[70] R. Strelitzki, J.A. Evans, A.J. Clarke, The influence of porosity and

pore size on the ultrasonic properties of bone investigated using aphantom material, Osteoporos Int. 7 (1997) 370–375.

[71] Y. Xia, W. Lin, Y.X. Qin, The influence of cortical end-plate onbroadband ultrasound attenuation measurements at the human

calcaneus using scanning confocal ultrasound, The Journal of theAcoustical Society of America 118 (2005) 1801–1807.