Osteoporosis Int (1997) 7:7-22
© 1997European Foundation for Osteoporosisand the National OsteoporosisFoundation
Review Article
The Role of Ultrasound in the Assessment of Osteoporosis: A Review
C. F. N j e h 1, C. M. Boivin 1 and C. M. Langton 2
1MedicalPhysicsDepartment, UniversityHospital Birmingham NHS Trust, Birmingham, UK; 2Centre for MetabolicBone Disease, Hull
Royal Infirmary, Royal Hull HospitalsNHS Trust and Universityof Hull, Hull, UK
Abstract. Osteoporosis is now being recognized as a
"silent epidemic" and there is an increasing need to
improve its diagnosis and management. Quantitative
ultrasound (QUS) measurement [broadband ultra-
sound attenuation (BUA) and velocity] is emerging as
an alternative to photon absorptiometry techniques in
the assessment of osteoporosis. The fundamental princi-
ples governing ultrasound measurements are discussed,
and some of the commercially available clinical systems
are reviewed, particularly in relation to data acquisition
methods. A review of the published in vivo and in vitro
data is presented. The general consensus is that
ultrasound seems to provide structural information in
addition to density. The diagnostic sensitivity of
ultrasound measurement of the calcaneus in the pre-
diction of hip fracture has been shown by recent large
prospective studies to be similar to hip bone mineral
density (BMD) measured with dual-energy X-ray
absorptiometry (DXA) and superior to spine BMD.
Ultrasound has also been shown to correlate better with
the type of hip fracture (intertrochanteric or cervical)
than BMD and to provide comparable diagnostic sen-
sitivity to spine BMD in vertebral fractures. It has also
been observed that combining the results of both
ultrasound and D X A BMD significantly improved hip
fracture prediction. Areas where further research is
required are identified.
Keywords: Bone mineral density (BMD); Bone
structure; Broadband ultrasound attenuation (BUA);
Fracture; Osteoporosis; Speed of sound (SOS); Quanti-
tative ultrasound (QUS)
Correspondence and offprint requests to: C. F. Njeh, Bone Densito-
metry Services, Medical Physics Department, Queen Elizabeth
Hospital, Birmingham B15 2TH, UK. Tel: + 44 (121) 627 2591. Fax:
+ 44 121 627 2386.
Osteoporosis
International
Introduction
Osteoporosis has recently been recognized as a silent
epidemic and much effort is being directed towards its
diagnosis and management. Osteoporosis results from a
period of asymptomatic bone loss and hence reduced
bone strength, predominantly in cancellous bone. The
most important consequence is an increase in fractures,
which may occur either spontaneously or after minimal
trauma, when loss of bone is sufficient to cause mechani-
cal weakness [1]. At present the clinical assessment of
osteoporosis relies mainly on bone mineral density
(BMD) measurements. Several precise and accurate
methods of BMD measurement have been developed
including single gamma photon absorptiometry (SPA),
dual gamma photon absorptiometry (DPA), dual-
energy X-ray absorptiometry (DXA) and quantitative
X-ray computed tomography (QCT). These methods
have been reviewed elsewhere [2-6]. All these
techniques use ionizing radiation and are relatively
expensive and bulky.
BMD explains about 70-75% of the variance in
strength (the ability to withstand an applied load), while
the remaining variance could be due to cumulative and
synergistic effects of other factors such as bone micro-
structure, architecture, measurement artefacts and the
state of remodelling [7,8]. Bone architecture refers to
the three-dimensional arrangement of trabecular struts.
It can be quantified in terms of porosity (volume
fraction), connectivity (degree of connection of trabe-
cular struts) and anisotropy (orientation of struts).
Remodelling refers to the continuous formation and
destruction of bone by bone cells (osteoclasts and
osteoblasts) [9]. Although osteoporosis results from a
complex incompletely understood set of physiological
and biochemical processes, the clinical manifestation
(fracture or deformation) is purely mechanical.

Ultrasound is a travelling mechanical vibration and
the mechanical properties of the medium progressively
alter the shape, intensity (energy per second per unit
area) and speed of the propagating wave. Ultrasound
therefore represents an alternative to conventional
absorptiometry which seems to provide structural infor-
mation in addition to density [10,11].
There is a growing interest in the use of ultrasound
measurements for the detection and management of
osteoporosis [12,13]. This article presents a review of
published data on the role of ultrasound in the assess-
ment of osteoporosis. The references quoted are
intended to be representative rather than exhaustive.
Attention will be drawn to sources of error and variation
in accuracy reported from different centres and to the
different measuring protocols for velocity used by the
various commercial systems. The review starts with a
brief introduction to the physics of ultrasound and its
mode of interaction with porous materials such as
cancellous bone. The authors discuss the clinical role of
ultrasound, drawing a conclusion about its future and
suggesting areas in which further research is urgently
needed.
Ultrasound Characterization of Bone
The ability of most physical methods of diagnosis to
provide information on the properties of a particular
medium depends on the way in which the interrogator
(an ultrasound wave in this case) is modified by the
medium. In ultrasonic propagation through bone, both
the velocity of transmission and the amplitude are
affected by the medium. The bone tissue may therefore
be characterized in terms of ultrasound velocity and
ultrasound attenuation.
Velocity
The velocity of an ultrasound wave depends on the
material properties of the medium through which it is
propagating and its mode of propagation. Longitudinal
waves are the common mode of ultrasound propagation
through tissue. Materials with a high elastic modulus,
such as bone, can support additional propagation modes
such as shear waves. In a straight uniform bar with a
radius that is small compared with the wavelength of the
propagating ultrasound wave, the velocity (V) can be
related to the mechanical properties of the medium as
follows [14]:
v = /E/p (i)
where p is the density of the medium and E is the
Young's modulus (a measure of resistance to deforma-
tion). This equation does not strictly apply to
anisotropic, heterogeneous and dispersive (frequency-
dependent) materials such as bone. A more elaborate
equation exists, but Eq. i provides a first-order estimate
c. F. Njeh et al.
of the relationship between the mechanical properties
of bone and velocity [15].
The experimental methods of velocity measurement
have been reviewed by Breazeale et al. [16]. Clinical
velocity measurements may be achieved using either
pulse-echo (reflection) or transmission techniques using
single, large-diameter piezoelectric transducers. The
pulse-echo technique utilizes a single transducer to
transmit and receive the signal. The generated
ultrasound pulse travels through the sample and is
reflected at an interface to be detected by the same
transducer. In the transmission method, one transducer
acts as transmitter and the second as receiver. For
application to bone, the transmission technique is pre-
ferred because of bone's highly attenuating nature. The
velocity of ultrasound in the medium can be calculated
by dividing the propagation distance by the transit time.
In the clinical application of ultrasound velocity
measurements, there is no convention for the use of
terms, or for the calculation protocol for velocity. For
example, speed of sound (SOS), velocity of sound,
apparent speed of ultrasound, apparent velocity of
ultrasound (AVU) and ultrasound transmission velocity
(UTV) all refer to the same generic ultrasound measur-
ement. For the calcaneus, three different methods of
calculating velocity have been utilized, resulting in the
limb (heel) velocity (catcaneus ptus soft tissue), bone
velocity (calcaneus only) and time of flight velocity
(TOF) with a fixed transducer separation (Fig. 1).
The TOF method measures the transit time through
water, with and without the sample, using fixed trans-
ducer separation. Let x be the limb (heel) thickness
including soft tissue and Xb be the calcaneus thickness
without overlying soft tissue, tx and tb are the corres-
ponding transit times through x and xb respectively. Let
the soft tissue thicknesses and transit time through them
be s l , s2 and tl, t2 respectively. If Vw is the velocity of
ultrasound in water and At is the difference in transit
time with and without the sample in the TOF, then:
Limb velocity - tx
x
(2)
Bone velocity = X b _ x - - ( s i + s 2 )
tb tx--(t,+t2) (3)
= Vwx
TOF velocity x - (At Vw) (4)
In clinical measurements the TOF velocity method
assumes a constant heel thickness and therefore the
velocity measured is dependent upon heel width
[17,18]. The three velocity calculations yield slightly
different values but correlate strongly with each other
(r = 0.83-0.98) [19]. Miller et al. [19], using the CUBA-
Research machine, showed that TOF had the optimum
precision [coefficient of variation (CV) = 0.7%] but the
smallest dynamic range, whilst bone velocity (CV =
2.7%) had the largest dynamic range - a factor that
enhances its sensitivity. Precision expressed as CV is
defined as 100 x Standard deviation/Mean. Miller et al.
Ultrasound in the Assessment of Osteoporosis

Transmitting
Transducer
x&
Bone
Coupling
~TRre~eiving
nsducer
Gel
9

Soft Tissue
x~ t v
(a) Contact Method

Temperature
B o n / Regulated Water
Bath
Transmitting
Transducer Receiving
transducer

t Soft Tissue
x v

(b) Substitution Method

Transmitter Receiver
\
..... :.\~........
_ ~
2. ....Soft
./..........
Tissue
() x,t Tiblal B o n e )

(c) Tibial Method

Fig. 1. Schematic representation of different methods of ultrasound velocity measurement.

[19] recommended that TOF velocity should be
performed for immersion techniques but limb (heel)
velocity for contact measurements.
Attenuation
As an ultrasound beam interrogates a medium some of
its energy is lost, and this phenomenon is known as
attenuation. The intensity of a plane wave (wave made
of parallel wavelets) propagating in the y-direction
decreases with distance as
the medium via conversion to heat, due mainly to
internal molecular friction. Scattering effects occur
when particles absorb part of the ultrasound energy and
re-radiate it in all directions. The amount of scattering
depends primarily on the ratio of ultrasound wavelength
to the size of the scattering particle and on the acoustic
impedance of the scattering particle [21].
In the frequency range 0.1-1 MHz that is most useful
for bone characterization, the total attenuation is
approximately linearly proportional to frequency [22]:
/.t(f) = a f (6)

/~ = Ioe- ~ Y (5) where o~ is the slope of attenuation as a function of

where /~(f) is the frequency (f)-dependent intensity
attenuation coefficient (dB/cm), Io is the incident inten-
sity and Iy is the intensity at a distance y.
Factors contributing to attenuation include beam
spreading (diffraction), scattering, absorption and
mode conversion [20]. The predominant attenuation
mechanism in cancellous bone is scattering, while
absorption predominates in cortical bone.
Absorption is the dissipation of ultrasonic energy in
frequency in dB/MHz/cm. In clinical practice, this has
become known as broadband ultrasound attenuation
(BUA). In contrast to velocity, no theoretical relation-
ship between ultrasound attenuation and the mechani-
cal properties of cancellous bone has been established.
Experimental methods of measuring attenuation
have been reviewed by Breazeale et al. [16]. Like
velocity, attenuation can be measured in both pulse-
echo and transmission modes. For highly attenuating
materials such as cancellous bone, the transmission
10
mode is more practical since the acoustic wave passes
through the material only once. The clinical systems
that measure B U A have adopted the substitution
method described by Langton et al. [22]. The increase in
attenuation as a function of frequency is measured by
comparing the amplitude spectrum for a reference
material (degassed water using a surfactant) with that
for the measured sample (Fig. 2). This can be mathema-
tically derived as follows. Expressing Eq. 5 in terms of
the amplitude spectrum, the amplitude spectrum Ab(f)
of a pulse through cancellous bone can be expressed as a
function of the reference pulse Aw(f) through water:
Ab(f) = Aw(fi)e-~')YTtbTbt
where Ttb and Tbt are the amplitude transmission
coefficients from soft tissue to bone and from bone to
soft tissue, assuming the transmission coefficient
between water and soft tissue to be unity. Hence Eq. 7
can be solved for the attenuation coefficients (dB/cm) as
/2(f) 8.686 In [ Aw(y')
= T ~Ab(f)] + In (TtbTbt)
The slope of attenuation (BUA) in dB/MHz/cm is
given by the linear regression of the spectral amplitude
difference in Eq. 8. Clinical systems generally do not
normalize attenuation to calcaneal thickness and there-
fore report ax rather than o~. Due to the anatomical
variations in human subjects it would seem reasonable
to attempt to normalize for heel thickness. The effect of
normalization for calcaneal thickness has been investi-
gated both in vitro [15,18,23-25] and in vivo [23,26].
Using a Walker Sonix UBA575, Wu et al. [23] con-
cluded that B U A results depended on bone thickness in
a non-linear fashion and a simple normalization of bone
thickness by specifying the results in units of dB/MHz/
cm is inadequate. Kotzki et al. [18] found no significant
correlation (r = 0.004, p = 0.93) between heel width
and BUA. Contrary to these findings, Serpe and Rho
[24] and Njeh [15] reported a linear relationship
between B U A and sample thickness in in vitro bovine
Reference BUA= Slope
< been suggested that the phalanges could be used to
Bone
a FrequencyMHz Frequency MHz
Fig. 2. Description of B U A m e a s u r e m e n t , a the amplitude spectra for
a sample and reference material are compared, resulting in b attenu-
ation as a function of frequency. B U A is the slope of the regression
line.
C . F . Njeh et al.
and phantom studies respectively. These two in vitro
studies used thickness ranges of 5-35 mm and 6-40 mm
respectively which are different from the in vivo biologi-
cal range such as 38-67 mm observed by Kotzki et al.
[18]. On the other hand Blake et al. [26] reported that
normalization for bone width marginally increased the
area under the receiver operating characteristic (ROC)
curve (see Fig. 3), but the difference was not statistically
significant. In a study of 28 postmenopausal women Wu
et al. [23] observed only + 3 mm variability in heel
width. This led them to the conclusion that the impact of
bone size on B U A in the clinical setting should be small,
except where there is a marked variability of calcaneal
(7) bone width as in paediatric studies. However, a
reduction in the precision error associated with thick-
ness measurement might improve the diagnostic sensiti-
vity of normalized over non-normalized BUA.
Clinical U l t r a s o u n d B o n e M e a s u r e m e n t
Systems
(8) Ultrasound instruments have been commercially avail-
able only in the last few years. They include: UBA575+
(Hologic, Waltham, MA), Achilles (Lunar, Madison,
WI), CUBAClinical (McCue, Winchester, UK),
SoundScan 2000 (Myriad, Rehovot, Israel) and DBM
Sonic (IGEA, Italy). These systems measure either
ultrasound velocity, attenuation or both. Most
manufacturers have quoted precision values for the
parameters measured. This is usually quoted as the CV
from repeated measurements (Table 1).
The calcaneus is the most popular measurement site
for several reasons. First, it is 90% cancellous bone,
which due to its high surface-to-volume ratio has
approximately 8 times the metabolic turnover rate of
cortical bone. Hence cancellous bone will manifest bone
metabolic changes before cortical bone [27]. The calca-
neus is also easily accessible and the medio-lateral
surfaces are fairly flat and parallel, thus reducing repo-
sitioning errors. The choice of the calcaneus as the test
site has been supported by Wasnich et al. [28] and Black
et al. [29], who reported that the catcaneus appeared to
be the optimal BMD measurement site for routine
screening of perimenopausal women to predict the risk
of all fractures.
The SoundScan 2000 measures ultrasound velocity
along the cortical bone of the tibia. The manufacturers
claim that cortical bone is the predominant contributor
to fracture resistance [30]. The use of the phalanges,
with approximately equal contributions of cortical and
cancellous bone, has also been suggested [31]. It has
rofic
monitor rheumatoid arthritis patients [32]. There is
b little published data to validate the clinical suitability of
the tibia and the phalanges in terms of diagnostic
sensitivity for distinguishing normal from osteoporotic
patients; however, these machines are relative new-
comers and must await further assessment.
Ultrasound in the Assessmentof Osteoporosis
A brief description will be given of the commercial
systems mentioned above.
UBA575 Walker Sonix (Hologic)
The Walker Sonix UBA575 was the first commercial
system for the assessment of bone status. From the
initial work of Langton et al. [22] a prototype instru-
ment UBA1001 was developed by Walker Sonix Inc.
This system had a variable-frequency tone burst gener-
ator producing a sequence of short bursts of single-
frequency ultrasound rather than a single broadband
pulse. The frequency range of 200-1000 kHz was pro-
duced in 28-kHz steps. The UBA1001 provided only
B U A measurements carried out in a room-temperature
water bath between two fixed 25-mm-diameter trans-
ducers, both resonant at 1 MHz [33].
The UBA575 model introduced the refinements of
rectilinear scanning of the transducers and a pneumatic
foot restraint to improve reproducibility. The rectilinear
scan was made up of nine points in a 3 x 3 grid in a
region 22 x 22 mm. The transducers were changed to 18
mm diameter with 500 kHz resonant (nominal) fre-
quency and signal amplitudes registered at 12 equally
spaced intervals between 200 and 600 kHz.
A reference trace is obtained with degassed water,
then the subiect's heel is immersed and allowed to
stabilize for approximately 3 min. The calculation of
B U A of each point is as previously described, the final
result being obtained via a nine-point averaging
sequence. The UBA575 has recently been upgraded to
the UBA575+ with the capability of measuring velocity
by the substitution method. A pulse-echo technique is
used to determine the width of the calcaneus [34]. B U A
is now calculated using a fast Fourier transform method
(FFr) rather than the tone burst technique. The system
is now being marketed by the Hologic Corporation.
Achilles (Lunar)
Similar to the Walker Sonix, the Achilles uses a water
bath but maintained at a constant 37 °C, since both
velocity and B U A are temperature dependent [15, 35].
Two unfocused 25-mm-diameter transducers (originally
focused) are mounted coaxially at a fixed separation of
approximately 95 mm [36].
Data are collected in the time domain, thus enabling
both velocity and B U A to be measured. Velocity is
calculated using the TOF substitution method (Eq. 4),
with the transit time measured from the zero-crossing
point of the first negative slope of the received signal
trace. An assumption that all subjects have the same
heel width of 40 mm is used in the velocity calculation.
An FFT algorithm is used to calculate the amplitude
spectrum. B U A is then calculated by the same sub-
stitution method described previously but not norma-
lized for heel width.
Lunar have introduced an index called stiffness (not
11
related to mechanical stiffness), defined as a combi-
nation of normalized velocity and BUA:
Stiffness = 0.67 x B U A + 0.28 x S O S - 420
Stiffness is claimed to improve the standardized coeffi-
cient of variation of velocity or B U A alone [37, 38].
CUBA Clinical (McCue)
A detailed description of the mode of operation of the
CUBAClinical is given by Langton et al. [39]. Two 19-
mm-diameter broadband ultrasound transducers are
positioned in direct contact with the subject's skin over
the calcaneus with a constant pressure. The clinical
version has coupling silicone pads which assist in accom-
modating variation in heel shape. It is claimed that to
improve precision, an iterative algorithm ensures
consistency of coupling and positioning by requiring the
complete removal of the heel from the system, recoup-
ling and repositioning.
The time of ultrasound propagation through the heel
(transit time) is automatically measured from the start
of the transmitted wave to the start of the leading edge
of the received pulse. The heel width is also auto-
matically measured and the velocity thus calculated by
dividing the heel width by the transit time.
B U A is calculated via an FFT algorithm. The calcu-
lated amplitude spectrum for a subject's heel is com-
pared with a reference spectrum stored within the
system (see Eq. 8). B U A values are not normalized for
heel thickness in the clinical system. There is a
CUBAResearch version that allows the user to norma-
lize B U A for heel thickness and permits varying trans-
ducer arrangements and frequency ranges.
SoundScan 2000 (Myriad)
The SoundScan 2000 system differs from the three
previously mentioned in that it measures velocity along
a defined and fixed longitudinal distance of the cortical
layer of the mid-tibia parallel to its long axis (Fig. lc).
The mid-tibia was chosen because of its long straight
and smooth surface. This site is defined as the mid-point
between the distal apex of the medial malleolus and the
distal aspect of the patella [40]. The transducers are
coupled to the anterior surface of the tibia through
standard ultrasound gel. The transit time of a 250-kHz
pulse along a defined 50 mm distance is measured. A
single-unit ultrasound probe houses all the components,
together with a proprietary array of ultrasound trans-
ducers designed to eliminate the effect of soft tissue on
velocity. The probe is moved back and forth across the
tibial surface and a minimum of 150-200 discrete velo-
city readings are recorded. The resultant velocity is a
computed average of the five highest readings. It is
12 C. F. Njeh et al.

claimed that velocity is independent of overlying soft peaks, peak regression and the trend of velocities in the
tissue. four fingers.

DBM Sonic 1200 (IGEA) Comparison

The DBM Sonic 1200 uses the transmission technique to
measure amplitude-dependent ultrasound velocity
through the proximal phalanges of the fingers [31]. Two
16-ram-diameter, 1.25-MHz transducers are assembled
on a high-precision calliper (+ 0.02 ram) that measures
the distance between the probes. The probes are
positioned on the medio-lateral phalangeal surfaces
using the phalanx head as reference point. Coupling is
achieved by means of standard ultrasound gel. The TOF
is defined as the time from emitted pulse to received
signal that is above a predetermined amplitude value.
When a normal bone is tested the amplitude of the first
signal received is above the predetermined threshold,
but for osteoporotic bone significant attenuation occurs
and the amplitude of the first signal is not enough to
trigger the reading. The velocity thus measured is
amplitude related. This enables the differences in SOS
measured between normal and osteoporotic bone to be
magnified. The probes are rotated until the best signal
(defined in terms of number of peaks and the amplitude
of the peaks) is recorded on the screen. Measurements
are carried out on each of the four phalanges and the
results averaged. The proximal phalanges were chosen
because of their cortical and cancellous bone com-
position. The medio-lateral surfaces are approximately
parallel, hence reducing ultrasound scattering.
The manufacturers are also working on improving the
system's diagnostic sensitivity by generating what is
termed "trace score" analysis. This grades the transmit-
ted ultrasound signal in terms of a number of para-
meters which are affected by the status of the bone.
These include: number of peaks, relative amplitudes of
As described in the preceding sections the clinical
systems have differences in the diameter of transducer
used, the frequency range and the method of measuring
velocity, resulting in a disparity in the results obtained
and difference in the dynamic range of values.
Gluer et al. [41] carried out a study on a group of 30
women using both the Walker Sonix UBA575 and the
Lunar Achilles. They reported that the B U A readings
obtained on the two machines differed substantially and
that the correlation between UBA575 B U A and
Achilles B U A was markedly lower than correlations
between bone densitometers measuring the same site.
Therefore B U A results cannot accurately be extra-
polated from one device to the other. No comparative
assessment of velocity has been reported for these
clinical systems.
The measurement precision varies depending upon
the parameter measured, the site and the system. The
manufacturers' quoted values are presented in Table 1.
The short-term in vivo precision of the CUBAClinical
has been reported to be 3.1% for B U A [42]. A review of
recent reported data using the Achilles system has
shown an average precision of 1.7% for BUA, 0.4% for
SOS and 1.7% for stiffness [43]. The reported precision
for the UBA575 varies between 2.1% and 4.6% for
B U A (see Table 3). A comparative study by Ramalin-
gam et al. [44] found a B U A precision of 6.3%, 1.5%
and 4.1% for CUBAResearch (rather than CUBA-
Clinical), Achilles and UBA575 respectively, and a
velocity precision of 1.04% and 0.27% for CUBA-
Research and Achilles respectively. The limitation of
using the CV as a figure of merit is its failure to take

Table 1. Commercially available ultrasound systems with the manufacturer's quoted precision

Systems Parameter Site Precision Comment

measured (CV%)

Achilles BUA and TOF Calcaneus BUA (2.0) Fixed distance,

(Lunar) Velocity Velocity (0.5) water bath
CUBAClinical BUA and limb Calcaneus BUA (3.3) Contact method,
(McCue) velocity Velocity (0.3) gel coupling
DBM Sonic Limb velocity Proximal Velocity (0.5) Contact, amplitude
(IGEA) phalanges dependent velocity
Signet (Osteo- Limb velocity Patella Not available No longer
Technology) available
SoundScan 2000 Bone velocity Tibial cortex Velocity (0.3) Contact, soft
(Myriad) tissue correction
UBA575+ BUA and bone Calcaneus BUA (2.0-4.0) Fixed distance,
(Hologic) velocity Velocity (0.5) rectilinear scanning

BUA, broadband ultrasound attenuation; TOF, time of flight.

Ultrasound in the Assessment of Osteoporosis
account of the range of values that B U A and velocity
may take in clinical studies. For this reason standardized
CV (SCV) defined as the percentage CV divided by the
ratio of the range over the mean may be preferable.
Ramalingam et al. [44] reported B U A SCV to be 8.4%,
5.1% and 6.2% for CUBAResearch, Achilles and
UBA575 respectively.
Factors Affecting Precision
In vivo precision is affected by soft tissue, heel thick-
ness, coupling and repositioning error. Using the
Walker Sonix UBA1001, Evans et al. [45] carried out an
extensive study of the factors that might affect pre-
cision. These included: immersion time of foot in the
water bath, water depth, water temperature, concentra-
tion of detergents and various rotations of the foot.
They found that foot positioning was the major cause of
measurement imprecision in BUA. This is due to the
spatial variation of B U A in the calcaneus. They pro-
posed that the optimum measuring temperature of the
water bath was 32 + 2°C, because they demonstrated a
decrease in B U A with temperature above 34°C. The
thickness of the overlying soft tissue has been reported
using the Achilles system to affect only velocity and not
B U A [181. One might expect that for the UBA575+, in
which soft tissue corrections are made, the effect will be
negligible.
Review of In Vitro and In Vivo Studies
A number of techniques which use ultrasound in the
diagnosis of bone disease have been reported in the
literature over the years. They use either or both
ultrasound velocity and attenuation, and this section
reviews the application of these techniques to the
assessment of bone, both in vitro and in vivo. As a new
technique coming into a well-established field, the
questions to be asked are:
What is ultrasound measuring?
How does ultrasound compare with ionizing radiation
methods?
What is the role of ultrasound in the assessment of bone
pathology?
Velocity Measurements
Studies of ultrasonic velocities in bone for clinical
purposes have been reported for at least 30 years. Most
of the early reported data for bone velocity were for
cortical bone, measured at various anatomical positions
and studied both in vivo and in vitro [46-50]. TOF
measurements and pulse-echo techniques were the most
commonly used methods in these investigations.
13
In Vitro Velocity Measurement. Ultrasound velocity in
cancellous bone measured in vitro has been reported
[50-55]. Abendschein and Hyatt [46], applying Eq. 1,
found a close correlation between the mechanically and
ultrasonically determined elastic modulus of human and
bovine cortical bone. These findings have been sup-
ported by the work of Ashman et al. [51] and Turner
and Eich [52]. Evans and Tavakoli [53] demonstrated a
highly significant correlation (r = 0.85) between velo-
city and physical density of bovine femur. Recently
Njeh et at. [54] and Bouxsein et al. [55] have reported a
significant correlation (r = 0.82 and 0.71 respectively)
between ultrasound velocity and ultimate strength of
the human calcaneus.
In Vivo Velocity Measurement. There has been a
tendency to correlate ultrasound velocity with the estab-
lished ionizing radiation measurements of BMD. The
correlation coefficients have ranged from 0.34 to 0.72
(Table 2). These correlations, although significant, are
weak. It has been argued that this is due to velocity
being dependent upon other aspects of bone in addition
to density. On the other hand, if Eq. 1 holds, then the
relationship between velocity and density should be
dependent on the relationship between Young's modu-
lus and density. For example if E o~ 0 2, then V oc ~/P.
Therefore it would be more appropriate to look for a
power relationship than a simple linear regression.
Other interesting relationships which have been dis-
cussed include the work of Rubin et al. [56], who
demonstrated changes in velocity with immobilization
and sporting activity.
Velocity Diagnostic Sensitivity. The first full docu-
mentation of the clinical application of velocity in the
detection of osteoporotic fragility was reported by
Heaney et al. [57]. They found a significant difference
between what they termed apparent velocity of
ultrasound (apparent because no compensation is made
for the overlying soft tissue) at the patella in osteoporo-
tic subjects compared with normal subjects. From a
clinical viewpoint, the diagnostic value of a test can be
calculated using Z-scores, R O C curve analysis and odds
ratios (relative risk). The Z-score is a measure of the
difference between the patient's measurement and the
mean measurement of age- and sex-matched peers. The
R O C curves are generated from the sensitivity (mea-
sure of true positive) and specificity (measure of true
negative) of the method (Fig. 3). Heaney et al. [57],
using ROC curves, came to the conclusion that velocity
measurement was an equally sensitive indicator of bone
fragility in postmenopausal and osteoporotic women as
bone mineral content (BMC) of the spine measured
using DPA. They also found that the velocity generally
decreased with increasing age after menopause. Ross-
man et al. [58] demonstrated that for discrimination of
spinal osteopenia, the area under the R O C curve was
similar for radius BMD (0.88) measured by SPA and for
velocity (0.86) through the calcaneus.
Recent studies [59-66] have demonstrated that
14 C.F. Njeh et al.

Table 2. Published correlations between velocity and BMD at various sites in vivo and precision of velocity measurement

Reference Ultrasound site BMD site BMD method r value Precision(CV%) No. of subjects

115 Distal radius Distal radius SPA 0.68 0.3 313F

116 Patella Spine DXA 0.36 1.1 153M
Spine DXA 0.34 129F
117 Patella Spine DXA 0.41 1.3 64F
Spine QCT 0.37 20F
118 Tibia Spine DXA 0.50 0.25 307F
Femur DXA 0.47
Radius SPA 0.63
33 Phalanges Spine DXA 0.39 0.7 30M
Femur DXA 0.47
Hand DXA 0.57
102 Calcaneus Spine DXA 0.55 ns 1000F
Femur DXA 0.54 1000F
119 Calcaneus Spine DXA 0.48 ns 300F
Femur DXA 0.45 300F
38 Calcaneus Spine DXA 0.54 0.2 107F
Femur DXA 0.65 107F
120 Calcaneus Femur DXA 0.35 0.12 244F
Total body DXA 0.56 244F
34 Calcaneus Calcaneus DXA 0.66 0.46 64F
36 Calcaneus Calcaneus DPA 0.72 1.2 64F
58 Calcaneus Radius SPA 0.66 1.0 76F
Spine DPA 0.63
Femur DPA 0.52

SPA, single photon absorptiometry; DXA, dual-energy X-ray absorptiometry; QCT, quantitative computed tomography;
DPA, dual photon absorptiometry; ns, not stated; F, females; M, males.

ultrasound velocity discriminates between fracture and
control groups. Turner et al. [59] reported that velocity
measured at the calcaneus had sensitivity comparable to
femoral neck B M D and better than spine B M D in hip
fracture prediction. For hip fracture they reported an
area under the R O C curve of 0.85 for SOS, 0.78 for
femoral neck B M D and 0.53 for spine B M D . It has also
been reported that ultrasound velocity correlates m o r e
highly with the type of fracture than B M D by D X A ,
with significantly lower velocity (p = 0.001) for patients
with trochanteric fractures than for those with cervical
fractures [60]. Velocity measured at the calcaneus has
o man et al. [65] reported that, in women, velocity
o measured at the tibia was superior to patella velocity
5 statistically significant difference in tibia velocity
r~
100% Specificity 0%
Fig. 3. Receiver operating characteristic (ROC) curves. The diagonal
line represents discrimination by pure chance. Curves above and to
the left of the diagonal line represent increasingly good discrimi-
nation. The area under the curve (AUC) can be used as a measure of
the power of a test.
similar diagnostic sensitivity to spine B M D and femoral
neck B M D for distinguishing patients with vertebral
fractures from controls [59,61]. Wuster et al. [61]
reported an area under the R O C curve of 0.85 for SOS
and 0.79 for spine BMD. However, Gonnelli et al. [62]
observed that although SOS was an independent predic-
tor of vertebral fracture, B M D was a slightly better
predictor. Using odds ratios analysis, Stegman et al. [63]
demonstrated that apparent ultrasound velocity at the
patella was as sensitive as forearm SPA in estimating the
likelihood of low-trauma fracture. In a 2-year prospec-
tive study H e a n e y et al. [64] confirmed the ability of
ultrasound velocity at the patella to predict vertebral
fractures. For low-energy appendicular fractures Steg-
and radial B M D , but that there was no difference
between the three modalities in men. Contrary to
Stegman et al. [65], Orgee et al. [66] observed a
between normal individuals and those with vertebral
osteoporosis. It is thought that the discrepancies
between these two studies could be due to the severity
or definition of vertebral deformation. In most of the
above-reported studies ultrasound was still a significant
predictor of fracture after adjusting for B M D of the
fracture site.
The first prospective study of hip fractures measuring
Q U S and B M D reported by Hans et al. [67] observed
that velocity measured at the calcaneus had the same
diagnostic sensitivity as femoral neck B M D in predict-
ing hip fractures. The increased risk associated with a
Ultrasound in the Assessmentof Osteoporosis
decrease of i standard deviation was estimated as 2.0 for
both ultrasound velocity and femoral BMD.
Attenuation Measurement
Until 1984 the literature on the diagnostic potential of
ultrasonic attenuation was sparse. Garcia et al. [68] used
an immersion transmission method, excited by both
continuous wave and broadband pulse techniques. They
demonstrated a decrease in attenuation of bovine corti-
cal bone with decrease in mineral content. Cancellous
bone attenuation was investigated by Fry, and Barger
[50] and Smith et al. [69]. Their work, however, was
confined to the skull for the purpose of finding the
optimum parameters for trans-skull diagnostic imaging.
Significant interest developed in the application of
ultrasound attenuation to bone pathology after the
work of Langton et al. [22]. They demonstrated that the
attenuation of ultrasound over a frequency of 0.24).6
MHz (termed Broadband Ultrasound Attenuation,
BUA) was linear and the slope of this relationship could
differentiate between normal and osteoporotic subjects
(Fig. 2b).
BUA In Vitro Studies. The relationship between B U A
and physical density has also been investigated in vitro.
McCloskey et al. [70] reported measurements of B U A
and physical density on samples of calcaneus obtained
from cadavers and concluded that there was a high
correlation (r = 0.85). McKelvie et al. [71] obtained a
similar correlation (r = 0.83) between B U A and physi-
cal density of the calcaneus. Evans and Tavakoli [53]
found a poor correlation (r = 0.33) between B U A and
physical density of bovine femur. It has been suggested
by Langton et al. [72] that the high correlation between
B U A and physical density in the human as compared
with the bovine samples could be due to smaller struc-
tural variation in the human calcaneus. They used
"fabric" as an index of structure and observed only a
6.7% variation, whereas that of density was 29.4%.
Tavakoli and Evans [11] investigated the dependence of
velocity and B U A on BMC. Three bovine femur speci-
mens were successively demineralized by nitric acid.
They each showed a good correlation between B U A
and bone density (r -- 0.84-0.99), velocity measure-
ments showing a higher correlation of greater than 0.97.
In a recent study Bouxsein et al. [73] found that both
calcaneal BMD and B U A were highly correlated with
femoral failure load (r = 0.63 and 0.51, respectively).
This is the only study, to the authors' knowledge,
addressing the association between femoral failure load
and densitometric measurements at sites other than the
femur.
BUA In Vivo Studies. A series of in vivo comparative
studies of B U A and BMD have been reported. Table 3
shows some of the published correlations between B U A
and various densitometric techniques, which vary from r
= 0.32 to 0.87. It is worth noting that similar correlation
15
coefficients have been reported between BMD at differ-
ent sites. Ross et al. [74] reported a correlation of r =
0.70 between calcaneal and spine BMD, r = 0.49
between spine and metacarpal BMD and r = 0.43
between calcaneus B U A and spine BMD. Kleerekoper
et al. [75] reported correlations of r = 0.64 between
spine and femoral neck BMD and r = 0.57 between
radial and spine BMD.
The interpretation of the correlations between B U A
and absorptiometry techniques (Table 3) is complicated
by many factors, including: measurement site assessed
by B U A and bone densitometry techniques, effects of
different ultrasound systems used, the different modes
of interaction of ultrasound- and photon-based tech-
niques, age range or population group studied, and
precision and accuracy. In a study by Gluer et al. [76]
the region of interest scanned by BMD was refined to
approximate more closely the region measured by
BUA. This did not significantly change the magnitude
of correlations between BMD and BUA. The
ultrasound systems from different manufacturers vary in
the transducer diameter used, nominal frequency,
transducer separation, frequency range used to evaluate
B U A and mode of coupling of transducer to patients
(i.e. contact or water bath). These parameters intro-
duce other variables which may explain the variation in
correlations.
The difference in precision between B U A and ioni-
zation techniques has been thought to partially account
for the poor correlation [76]. The reported short-term
precision in vivo of B U A at the calcaneus varies from
1.6% to 10% (see Table 3), whereas precision of SPA
and DXA is approximately 1-2.8% [5]. However, in a
study by Gluer et al. [76], adjustment of raw correlation
for measurement error in B U A and BMD resulted in
little improvement in correlation (0.64 ~ 0.76). This
suggests that imprecision accounts for only a small
portion of the unexplained variability between these
two methods [76]. Brandenburger [77], using data from
the literature, has argued that the variation in cor-
relation is because ultrasound measures clinically rele-
vant properties of bone in addition to, and distinct from,
bone mass. This property, termed "bone quality", is
dependent on age and site of measurement.
There is now general agreement that ultrasound and
DXA measure different aspects of bone and a high
correlation between them should not be expected.
Therefore simple linear regression is not really an
appropriate way of relating these two modalities.
Hence, it is not surprising that studies such as that of
Faulkner et al. [78] that have set out to predict axial
BMD from calcaneal B U A have not been successful.
Attention is now directed towards the diagnostic accur-
acy of ultrasound rather than simple correlation with
BMD using statistical techniques such as ROC analysis,
Z-scores and odds ratios [79,80].
BUA Diagnostic Sensitivity. The ability of B U A to
discriminate between normal and osteoporotic subjects
has been reported in the literature (Table 4). The
16 C . F . Njeh et al.

Table 3. Published correlations between B U A of the calcaneus and BMD at various sites in vivo and precision of B U A measurement

Reference BMD site BMD method r value Precision (CV%) Mean age (SD) No. of subjects System

74 Calcaneus SXA 0.68 ns 74 (5,2) 650F UBA575

121 Calcaneus SPA 0.66 3.6 63.5 (5.2) 261F UBA575
76 Calcaneus SXA 0.65 2.8 43,5 33X UBA575
76 Calcaneus SXA 0.75 2.8 45.6 24F UBA575
34 Calcaneus DXA 0,73 2.8 51 (10) 64F UBA575+
36 Calcaneus DPA 0,56 10 30 42M E
74 Distal radius SXA 0.42 ns 74 (5.2) 650F UBA575
85 Distal radius SPA 0,77 3. I 48 61F UBA?
122 Distal radius SPA 0.45 2.1 ns 21F UBA1001
123 Distal radius SPA 0.80 3.9 ns 44X UBA1001
124 Distal radius QCT 0.66 4 ns 24X UBA1001
125 Distal radius SPA 0.80 3.9 ns 44X UBA1001
58 Radius shaft SPA 0.64 10 45 76F UBA1001
124 Radius shaft QCT 0.85 4 ns 44X UBA1001
122 Radius shaft SPA 0.64 4.6 ns 21F UBA1001
74 Spree DXA 0.43 ns 74 (5.2) 650F UBA575
87 Spree DXA 0.50 3.9 53.7 (14.3) 400X UBA575
102 Spree DXA 0.54 ns 57.4 (12.8) 1000F Achilles
121 Spree DXA 0.43 3,6 63.5 (5.2) 261F UBA575
126 Spree DXA 0.42 2.6 47 (1.41) 328F UBA575
127 Spine DXA 0.35 6 52 578F UBA575
86 Spine DXA 0.61 2.3 58.1 (1.3) 58F UBA575
128 Spme DXA 0,83 2.9 57 (10) 22F UBA575
85 Spme DPA 0.72 3.1 48 61F UBA?
58 Spree DPA 0.66 10 45 76F E
122 Spree QCT 0.45 4.6 ns 21F UBA100t
38 Spree DXA 0.55 1.4 24-77 107F Achilles
119 Spree DXA 0.54 ns 53 300F CUBA
84 Spree DPA 0.61 2.6 65.4 (1.6) 61F UBA?
87 Femoral neck DXA 0.52 3.9 53.7 (14.3) 400X UBA575
102 Femoral neck DXA 0.55 ns 57.4 (12.8) 1000F Achilles
121 Femoral neck DXA 0.43 3.6 63.5 (5,2) 261F UBA575
126 Femoral neck DXA 0.32 2.6 47 (1.41) 328F UBA575
127 Femoral neck DXA 0.40 6 52 578F UBA575
86 Femoral neck DXA 0.68 2.3 58.1 (1.3) 58F UBA575
128 Femoral neck DXA 0.87 2.9 57 (10) 22F UBA575
58 Femoral neck DPA 0,41 10 45 76F E
84 Femoral neck DPA 0.59 2.60 65.4 (1.6) 61F UBA?
120 Total body DXA 0.53 0.93 31-79 244F Achilles

F, female; X, mixed; ns, not stated; E, experimental apparatus; UBA?, model not stated.

differences reported between normal and osteoporotic
volunteers are statistically significant. Resch et al. [81]
found a significant difference in BUA between 37
Caucasian women (mean age 65 + i years) with one or
more atraumatic vertebral fractures and 23 healthy
women (mean age 63 + 2 years). Variants in age do not
account for this significance, although a negative cor-
relation between B U A and age has been reported
[34,82]. Damilakis et al. [83] demonstrated using multi-
ple regression analysis that the difference between
normal and osteoporotic patients was more disease
related (p<0.001) than age related (p<0.05).
The sensitivity/specificity for various fractures using
different levels of B U A has been reported by Baran et
al. [84] to be 80%/80% at 70 dB/MHz, by McCloskey et
al. [85] to be 81%/93% at 50 dB/MHz, by Agren et al.
[86] to be 76%/76% at 63 dB/MHz and by Funke et al.
[87] to be 85%/85% at 64 dB/MHz. From these studies
we may conclude that B U A measurements of the
calcaneus have a similar diagnostic accuracy to lumbar
spine BMD.
In a retrospective study of a group of 50 women who
had recently sustained a fracture of the hip, Stewart et
al. [88] demonstrated that B U A of the calcaneus had
diagnostic sensitivity comparable to hip BMD but
superior to spine BMD. The area under the ROC curve
(AUC) for B U A was 0.76, for BMD lumbar spine 0.61,
BMD neck of femur 0.62, BMD trochanter 0.66 and for
BMD Ward's triangle 0.66. The mean Z-scores in the
fracture patients were: B U A = -0.96, BMD spine =
-0.57, BMD neck of femur = -0.82, BMD trochanter
--- -1.01 and BMD Ward's triangle = -0.76. These
results have been supported by the work of Turner et al.
[59] and Schott et al. [60]. Turner et al. [59] found the
AUC to be 0.79 for BUA, 0.78 for BMD neck of femur
and 0.53 for BMD spine. They also demonstrated that
BUA adjusted for femoral neck BMD still discrimin-
ated between fracture and control groups. The poor
Ultrasound in the Assessmentof Osteoporosis 17

Table 4. Published BUA discriminationbetween normal and osteo- hip fracture risk in postmenopausal women. This study
porotic subjects did not measure BMD and the method of analysis did
not disclose whether B U A measurements were associ-
Reference No. of Selection Mean p value ated with fracture risk after controlling for age and other
subjects criteria BUA (SD) variables. There have been two large prospective
studies carried out by groups in France on 7598 very
84 N29 54.5 elderly women [67] and in the USA on 6183 elderly
O22 Osteopenia 40.3 <0.01 women [95]. Hans et al. [67] observed that for a 1
O10 FNF 32.2 <0.01
85 N24 Premenopausal 79.6 standard deviation decrease there was an age- and
N10 Perimenopausal 79.6 weight-adjusted relative hip fracture risk of 2.2 for
O21 VCF 55 <0.001 B U A compared with 2.0 for femoral neck BMD. After
94 N1341 50.9 (22.2) <0.001 adjusting B U A for femoral neck BMD, the logistic
073 FNF 40.3 (19.3) regressions showed that B U A was still a significant
81 N23 62.6 (2.9)
037 VCF 54 (2.2) <0.04 independent predictor of hip fractures. They also
86 N17 70.2 demonstrated that combining the results of ultrasound
O21 Withoutfracture 56.3 and BMD significantly improved the prediction of hip
O16 VCF 52.9 <0.001 fracture. Bauer et al. [95] found an adjusted relative risk
O4 FNF 43.0
82 N66 Premenopausal 85 (12) of 2.2, 2.5 and 3.0 for a 1 standard deviation decrease in
N27 Elderly 69 (12.5) calcaneal BUA, calcaneal BMD and femoral neck
033 VCF 55 BMD in hip fracture prediction. However contrary to
83 N74 66.5 (12) Hans et al. [67] they found that B U A was no longer
O19 Radiograph 48.8 (6) <0.001 significantly associated with hip fracture after adjusting
88 N50 67.98 (18.7) <0.0001
O50 FNF 49.90 for either hip or calcaneal BMD. It is worth noting that
87 N265 75 (14.3) these two studies were carried out using two different
O135 Osteopenia 59.7 (14.4) <0.001 ultrasound systems: the Lunar Achilles for the French
046 FNF 53.9 (12.9) study and the Walker Sonix UBA575 for the American
study. Recently Gluer et al. [96] reported that the
N, normal; O, osteoporotic; VCF, vertebral crush fracture; FNF, relative risk for hip fracture prediction varies between
femoral neck fracture. ultrasound systems. This could explain the dis-
crepancies reported between the two studies. The study
prediction of hip fracture by lumbar spine BMD may be population and the proportion of trochanteric and neck
because BMD is artificially but unpredictably elevated fractures could also contribute to the observed differ-
in the spine by aortic calcification and osteoarthritis, ences.
since these problems are common in the elderly [89]. The diagnostic sensitivity of B U A has been reported
Stewart et al. [90] have also addressed retrospectively for other skeletal diseases such as rheumatoid arthritis.
the relative abilities of B U A of the calcaneus and BMD Fordham et al. [97] found a significantly lower B U A for
of spine and hip to discriminate between vertebral rheumatoid arthritis patients compared with controls
deformation or fracture defined using radiographs. (p = <0.05); however, they failed to find any differ-
Three hundred women and 300 men of mean age 73.1 ence between the patients with nodal osteoarthritis and
and 65.2 years, respectively, were randomly selected control subjects. The inability to differentiate between
from the community. For discrimination of male verte- controls and osteoarthritis patients has been supported
bral deformations or fractures none of the measure- by the work of Nijs et al. [98]. Acotto et al. [99] found a
ments (BMD or BUA) achieved statistical significance. statistically significant lower B U A (p <0.005) in
However, for the women BMD was superior to BUA, patients with thyrotoxicosis compared with age-
with an A U C of 0.56 for B U A , 0.71 for BMD neck of matched controls. Jones et al. [100] also reported that
femur, 0.73 for BMD trochanter, 0.71 for BMD Ward's active and sedentary groups of volunteers could be
triangle and 0.72 for BMD spine. Gonnelli et al. [62] distinguished using B U A (p <0.001). Volunteers were
also observed that the relative risk of vertebral fracture included in the active group if they performed at least
was higher for low-spinal BMD than for BUA. Contrary 6 h per week of weight-bearing active exercise, and in
to these findings others have reported that B U A dis- the sedentary group if they did less than 1 h of weight-
criminated between subjects with vertebral fracture and bearing activity per fortnight. They also observed an
controls as effectively as spine BMD [59,74,87,91]. One increase in B U A (p <0.05) following a year of brisk
of the reason for discrepancies could be the lack of walking (16-18 km/week).
consensus as to what constitutes a vertebral fracture.
B U A was still a statistically significant predictor of
fracture after adjusting for spinal BMD [59,74,87,91]. Correlation Between Velocity and B UA
B U A has also been reported to be sensitive to other
fractures such as Colles' [92,93]. Very limited studies have addressed the relationship
The first prospective study for hip fractures by Porter between velocity and attenuation in cancellous bone in
et al. [94] demonstrated that calcaneal B U A predicts vivo. Rossman et al. [58] found the correlation between
18
B U A and SOS to be 0.53. They argued that the low
correlation was due either to SOS and B U A measuring
different aspects of bone or to the precision error of
each, particularly the relatively high precision error of
B U A (10%). However, Herd et al. [101] and Waud et
al. [34] have reported higher correlation coefficients of
0.75 and 0.74 respectively. A recent study by Rosenthall
et al. [102] using the Lunar Achilles on 1000 women
found a correlation coefficient of 0.64 between SOS and
BUA.
Lunar, the manufacturer of the Achilles, has derived
a value called "stiffness" (see the section on the Achilles
above) that combines velocity and BUA. This is sup-
posed to give an SCV closest to that of spine BMD.
Stiffness has been reported in a cross-sectional study by
Schott et al. [60] to give better sensitivity than femoral
BMD in hip fracture prediction. The stiffness Z-score
and T-score were reported to be - 1 . 0 and - 3 . 7 respec-
tively, compared with - 0 . 8 and - 2 . 8 for femoral neck
BMD.
Evidence of Structural Dependence of Bb¼ and
Velocity
Structure is one of the qualities of bone that ultrasound
is purported to measure - in contrast to DXA which
measures density only. Structure can be defined in terms
of connectivity, porosity and anisotropy. A few research
groups have attempted to demonstrate this structural
factor in ultrasound measurements. Langton et al. [103]
measured cubes of equine bone in three orthogonal
axes, and reported a significant variation between axes.
This variation with orientation has also been reported
by Nicholson et al. [104] in human vertebrae and Njeh
[15] in bovine femora. The density of an individual cube
is independent of the axis studied, so any variation in
ultrasound properties with orientation suggests a struc-
tural component. Tavakoli and Evans [105] used large
static loads to modify the porosity of bovine femur
samples, without changing the amount of bony tissue in
the ultrasound path. They concluded that the observed
changes in measured B U A were due to structure rather
than to the component material. Gluer et al. [10]
measured cubes in three orthogonal directions of can-
cellous bone cut from fresh bovine proximal radii. The
relative orientation of the trabeculae with respect to the
direction of the ultrasound beam was evaluated on high-
resolution conventional radiographs employing a semi-
quantitative "Alignment" score ranging from - 2 (for
perpendicular) to +2 (for parallel). They demonstrated
that Alignment showed significant association with
BUA, indicating that B U A depends on trabecular
orientation.
Other researchers such as Njeh [15] have used perme-
ability (a parameter that determines the rate of flow of
fluid or gas through a porous medium such as cancellous
bone) to quantify the structure of bovine and human
cancellous samples. Hans et al. [106] and Gluer et al.
[107] have studied the relationship between ultrasound
C. F. Njeh et al.
and histomorphometric parameters which reflect the
microarchitecture of bone. Gluer et al. [107] were able
to demonstrate significant association of bone structure
(connectivity and trabecular dimensions) and ultra-
sound parameters independent of BMD. On the other
hand Hans et al. [106] failed to show any significant
association between ultrasonic or densitometric para-
meters and microarchitecture. However, the sample
size and method of sample preparation could have
contributed to the poor association. The trabecular
thickness and separation that affect ultrasound trans-
mission are not the same in the human and bovine
samples. For example, the human calcaneal trabecular
thickness has been reported to be 85 + 21/~m [106] and
that of the bovine proximal radius to be 188 + 10#m
[107].
Future Studies and Prospects for QUS
Relatively poor precision in calcaneal B U A measure-
ment compared with DXA is a major cause of concern
in the clinical application of QUS. One of the causes of
poor precision is the heterogeneity of the calcaneus.
This gives rise to physical effects in the ultrasonic field
such as refraction and phase distortion resulting in
phase cancellation [108]. It has been reported that when
a large-aperture single-element piezoelectric transducer
is used as a receiver, phase cancellation artefacts can
cause overestimation of B U A values [109]. Poor
precision of measurements is therefore related to
transducer positioning [45,76]. Due to inter-individual
anatomical variation the site of measurement may vary
from one subject to another. The previously described
commercial systems have employed different
approaches to overcome this difficulty, but the precision
is still relatively poor. Laugier et al. [110] have proposed
a system that scans the whole calcaneus and produces
B U A images. Region of interest analysis could then be
used to select the optimal measuring site. Such a system
has been reported to provide in vivo precision varying
between 1.5% to 3.3% [111]. Further studies in the
diagnostic sensitivity of such an approach are war-
ranted.
As manufacturers use different transducer diameters
and frequency ranges, there is a need to investigate the
optimum frequency range and transducer size for repro-
ducible results. Cross-calibration equations for the
different systems are required so that measurements on
one could be interpolated to another.
There is a necessity for a better understanding of the
underlying mechanisms of ultrasound interaction with
cancellous bone. With different clinical systems measur-
ing different sites, cross-calibration is required. Such an
endeavour is one of the objectives of the European
Osteoporosis and Ultrasound Study (OPUS).
Phantoms are required which can be used in a number
of areas of quality control, particularly standardization
of different centres and equipment, clinical testing and
monitoring. Present manufacturers' phantoms do not
Ultrasound in the Assessment of Osteoporosis
have the heterogeneity observed in vivo. T h e r e is
therefore a need for an anthropomorphic ultrasound
phantom. Langton [112] has suggested an electronic
p h a n t o m for which the B U A values could be varied to
encompass the range observed clinically.
In vitro studies have indicated that prediction of bone
strength could be improved by combining velocity and
physical density, or velocity and B U A [54, 113]. In vivo
studies have also indicated that combining ultrasound
and D X A p a r a m e t e r s m a y improve diagnostic sensiti-
vity [67,114]. Therefore large prospective studies are
required to determine the best method of combining
velocity, B U A and B M D to improve fracture pre-
diction.
Few studies have addressed the usefulness of
ultrasound in monitoring treatment for osteoporosis, so
prospective studies are required.
Conclusions
The early scepticism about the role of ultrasound in
osteoporosis is decreasing as a result of the growing
body of evidence, particularly f r o m prospective studies,
that quantitative ultrasound is useful for the assessment
of bone. In vitro studies have demonstrated that both
ultrasound velocity and attenuation ( B U A ) seem to
give structural information. T h e question still not
answered is how to quantify the structural information.
There is therefore a need to determine which features of
ultrasound velocity and attenuation are related to bone
density and which reflect changes in architecture inde-
pendent of density and how they contribute to mechani-
cal competence. This can be achieved by establishing a
theoretical relationship between B U A , velocity, archi-
tecture and the mechanical properties of cancellous
bone.
Ultrasound p a r a m e t e r s ( B U A and velocity) at the
calcaneus are correlated with axial B M D to about the
same degree as peripheral B M D . H o w e v e r , ultrasound
of the calcaneus provides diagnostic sensitivity superior
to calcaneal B M D . It appears to have a sensitivity equal
to that of spine B M D for spine fracture, and equal to
that of femoral B M D and superior to spine or peripheral
B M D for hip fracture. Ultrasound is still a significant
predictor of fracture after adjusting for B M D . It is
better correlated with the type of hip fracture (trochan-
teric or cervical) than B M D . T h e r e is also evidence that
ultrasound combined with B M D could improve fracture
prediction.
T h e general consensus is that heel width has no
significant effect on B U A (due to small biological
variation) but has a slight effect on velocity measured by
the time of flight method. The role of velocity measured
at the calcaneus, patella, tibia and phalanges still needs
to be identified.
The absence of ionizing radiation, the portability of
the equipment and its relatively low cost makes
ultrasound assessment an attractive option for "screen-
ing" p r o g r a m m e s for fracture and studies of children,
19
and for research trials involving normal control groups.
It remains to be seen whether its primary role will be as a
replacement for D X A or as a c o m p l e m e n t a r y measure-
ment.
Acknowledgements. We are grateful to Dr W. D. Morgan for helpful
comments of the manuscript. C.F.N. is supported by the Birmingham
Medical Physics Trust Fund.
References
1. Stevenson JC, Whitehead MI. Postmenopausal osteoporosis.
BMJ 1982;285:585-8.
2, Tothill P. Methods of bone mineral measurements. Phys Med
Biol 1989;34:543-72.
3. Sire LH, van Doorn T. Radiographic measurement of bone
mineral: reviewing duN-energy X-ray absorptiometry. Aust
Phys Eng Sci Med 1995;18:65-80.
4. Alhava ME. Bone density measurement. Calcif Tissue Int 1991;
(suppl) 49:$21-3.
5. Lang P, Steiger P, Faulkner K, Gluer C, Genant HK. Current
techniques and recent developments in quantitative bone densit-
ometry. Radiol Clin North Am 1991;29:49-76.
6. Ostlere SJ, Gold RH. Osteoporosis and bone density measure-
ment methods. Clin Orthop 1991;271:14%63.
7. Kleerekoper M, Villaneuva AR, Stanciu J, Rao DS, Parfit AM.
The role of three-dimensional trabecular microstructure in the
pathogenesis of vertebral compression fracture. Calcif Tissue Int
1985;37:594-97.
8. Mosekitde L. Sex differences in age-related loss of vertebral
trabecular bone mass and structure: biomechanical conse-
quences. Bone 1989;10:425-32.
9. Marcus R. Understanding osteoporosis. West J Med
1991;155:53-60.
10. Gluer CC, Wu CY, Genant HK. Broadband attenuation signals
depend on trabecular orientation: an in-vitro study. Osteopor-
osis Int 1993;3:185-91.
11. Tavakoli MD, Evans JA. Dependence of the velocity and
attenuation in bone on the mineral content. Phys Med Biol
1991;36:1529-37.
12. Hans D, Schott AM, Meunier PJ. Ultrasonic assessment of
bone: a review. Eur J Med 1993;2:157.63.
13. Kaufman JJ, Einhorn TA. Perspectives: ultrasound assessment
of bone. J Bone Miner Res 1993;8:51%24.
14. Pain HJ. The physics of vibrations and waves. Chichester:
Wiley, 1985.
15. Njeh CF. The dependence of ultrasound velocity and attenu-
ation on the material properties of cancellous bone. PhD thesis,
Sheffield Hallam University, UK, 1995.
16. Breazeale MA, Cantrelt JH, Heyman JS. Ultrasonic wave
velocity and attenuation measurements. Methods Exp Phy
1981;19:67-135.
17. Hans D, Schott AM, Arlot ME, Sornay E, Delmas PD, Meunier
PJ. Influence of anthropometric parameters on ultrasound mea-
surements of os calcis. Osteoporosis Int t995;5:371-6.
18. Kotzki PO, Buyck D, Hans D, Thomas E, Bonnel F, Favier F,
Meunier PJ, Rossi M. Influence of fat on ultrasound measure-
ments of the os calcis. Catcif Tissue Int 1994;54:91-5.
19. Miller CG, Herd RJM, Ramalingam T, Fogelman I Blake GM.
Ultrasonic velocity measurements through the calcaneus: which
velocity should be measured? Osteoporosis Int 1993;3:31-5.
20. Bamber JC, Tristam M. Diagnostic ultrasound. In: Webb S,
editor. The physics of medical imaging. Bristol: Adam Hilger,
1988:31%86.
21. Kinsler LE, Frey AR, Coppens AB, Sanders JV. Fundamentals
of acoustics. New York: Wiley, 1992.
22. Langton CM, Palmer SB, Porter RW. The measurement of
broadband ultrasonic attenuation in cancellous bone. Eng Med
1984;13:89-91.
23. Wu CY, Gluer CC, Jergas M, Bendavid E, Genant HK. The
20
impact of bone size on broadband ultrasound attenuation. Bone
1995;16:137-4t.
24. Serpe LJ, Rho J. Broadband ultrasound attenuation value
dependence on bone width in vitro. Phys Med Biol 1996;41:19%
202.
25. Bouxsein ML, Radloff SE, Toledano TR, Hayes WC. Calcaneal
ultrasound measurements are moderately correlated with trabe-
cular bone density and independent of foot geometry. J Bone
Miner Res 1994;9(Suppl 1): $208.
26. Blake GM, Herd RJM, Miller CG. Should broadband ultrasonic
attenuation be normalized for the width of the calcaneus? Br J
Radiol 1994;67:1206-9.
27. Vogel JM, Wasnich RD, Ross PD. The clinical relevance of
ealcaneus bone mineral measuremnets: a review. Bone Miner
1988;5:35-58.
28. Wasnich RD, Ross PD, Heilbrun LK, Vogel JM. Selection of
the optimal skeletal site for fracture prediction. Clin Orthop
1987;216:262-9.
29. Black DM, Cummings SR, Genant HK, Nevitt MC, Palermo L,
Browner W. Axial and appendicular bone density predict
fractures in older women. J Bone Miner Res 1992;7:633-8.
30. Simkin A. Ultrasonic velocity along the tibial shaft as a diagnos-
tic tool for osteoporosis and a predictor of hip fractures. Myriad
Ultrasound Systems Ltd, 1993.
31. Mele R. Determinazione delle caratteristiche strutturali densito-
metriche del tessuto osseo mediante ultrasuoni. Atti della
Societa Emiliana Romagnola Triveneta di Ortopedia e Trauma-
tologia 1993;35(1):1-4.
32. Boivin CM, Njeh CF, Bulmer N, Lilley J, Delvin J, Emery P.
Finger ultrasound in assessment of rheumatoid arthritis. In
Zhonghot/ L e t al., editors. Advances in osteoporosis. Inter-
national Academic Publishers, 1995:392-3.
33. Truscott JG, Simpson M, Stewart SP, et al. Bone ultrasonic
attenuation in women: reproducibility, normal variation and
comparison with photon absorptiometry. Clin Phys Physiol
Meas 1992;13:29-36.
34. Wand CE, Lew R, Baran DT. The relationship between
ultrasound and densitometric measurements of bone mass at the
calcaneus in women. Calcif Tissue Int 1992;51:415-8.
35. Evans JA, Tavakoli MB. Temperature and direction depen-
dence of the attenuation and velocity of ultrasound in cancellous
and cortical bone. In: Ring EFG, Bhalla AK editors. Current
research in osteoporosis and bone measurement II. Bath: British
Institute of Radiology, 1992.
36. Zagzebski J, Rossman P, Mesina C, Mazess R, Madsen E.
Ultrasonic transmission measurements through the os calcis.
Calcif Tissue Int 1991 ;49:10%11.
37. Mazess R' Trempe J' Barden H" Ultras°und measurement °f the
os calcis. Calcif Tissue Int 1993;52:S165.
38. Lees B, Stevenson JC. Preliminary evaluation of a new
ultrasound bone densitometer. Calcif Tissue Int 1993;53:14%52.
39. Langton CM, Ali AV, Riggs CM, Evans GP, Bonfield W. A
contact method for the assessment of ultrasonic velocity and
broadband attenuation in cortical and cancellous bone. Clin
Physiol Meas 1990;11:243-49.
40. Orgee JM, Foster H, McCloskey EV, Khan S, Coombes G,
Kanis JA. A precise method for the assessment of tibial
ultrasound velocity. Osteoporosis Int 1996;6:1-7.
41. Gluer CC, Wu C, Genant HK. Disparity of different BUA
approaches. Calcif Tissue Int 1993;52:$171.
42. Graafmans WC, Lips P, Lingen AV, Bouter LM. Ultrasound
measurements in the calcaneus: reproducibility and its relation
with bone mineral density. In: Ring EFJ, Elvins DM, Bhalla
AK, editors. Current research in osteoporosis and bone mineral
measurements III. British Institute of Radiology, 1994:19.
43. Mazess R, Morris R, Trempe J. Recent advances in ultrasound
densitometry: the Achilles Plus. Industry Forum, l l t h inter-
national bone densitomerty workshop, Oregon, Sept 24-28,
1995.
44. Ramalingam T, Herd RJM, Lees B, Blake GM, Stevenson JC,
Miller CG, Fogelman I. A comparison of three commercial bone
ultrasound scanners (abstract). Calcif Tissue Int 1993;52:170.
45. Evans WD, Jones EA, Owen GM. Factors affecting the in vivo
C. F. Njeh et al.
precision of broadband ultrasonic attenuation. Phys Med Biol
1995;40:137--51.
46. Abendschein W, Hyatt GW. Ultrasonics and selected physical
properties of bone. Clin Orthop 1970;69:294-301.
47. Behari J, Singh S. Ultrasound propagation in in vivo bone.
Ultrasonics 1981 ;81:87-90.
48. Andre MP, Craven JD, Greenfield MA. Measurement of the
velocity of ultrasound in the human femur in vivo. Med Phys
1980;7:324-30.
49. Greenfield MA, Craven JD, Huddleston A, Kehrer ML,
Wishko D, Stern R. Measurement of the velocity of ultrasound
in human cortical bone in vivo. Radiology 1981;138:701-10.
50. Fry FJ, Barger JE. Acoustical properties of human skull. J
Acoust Soc Am 1978;63:1576-90.
51. Ashman RB, Corin JD, Turner CH. Elastic properties of
cancellous bone measurement by an ultrasonic technique. J
Biomech 1987;20:979-86.
52. Turner CH, Eich M. Ultrasonic velocity as a predictor of
strength in bovine cancellous bone. Calcif Tissue Int
1991 ;49:116-9.
53. Evans JA, Tavakoli MB. Ultrasonic attenuation and velocity in
bone. Phys Med Biol 1990;35:138%96.
54. Njeh CF, Hodgskinson R, Langton CM. Determination of bone
strength from ultrasonic velocity and broadband ultrasound
attenuation. Br J Radiol 1995;68:789.
55. Bouxsein ML, Radloff SE, Hayes WC. Quantitative ultrasound
of the calcaneus reflects trabecular bone strength, modulus and
morphology. J Bone Miner Res 1995; 10(Suppl 1):$175.
56. Rubin CT, Pratt GW, Porter AL, Lanyon LE, Poss R. The use
of ultrasound in vivo to determine acute change in the mechani-
cal properties of bone following intense physical activity. J
Biomech 1987;20:723-7.
57. Heaney RP, Avioli LV, Chestnut CH, Lappe J, Rescker RR,
Brandenburger GH. Osteoporotic bone fragility, detection by
ultrasound transmission velocity. JAMA 1989;261:298690.
58. Rossman P, Zagzebski J, Mesina C, Sorenson J, Mazess R.
Comparison of speed of sound and ultrasound attenuation in the
os calcis to bone density of the radius, femur and lumbar spine.
Clin Physiol Meas 1989;10:353-60.
59. Turner CH, Peacock M, Timmerman L, Neal JM, Johnston CC
Jr. Calcaneal ultrasonic measurements discriminate hip fractures
independently of bone mass. Osteoporosis Int 1995;5 130-5.
60. Schott AM, Weill-Engerer S, Hans D, Duboeuf F, Delmas PD,
Meunier PJ. Ultrasound discriminates patients with hip fracture
equally well as dual-energy X-ray absorptiometry and indepen-
dently of bone mineral density. J Bone Miner Res 1995;10:243-
9.
61. Wuster C, Paetzold W, Scheidt-Nave C, Brandt K, Ziegler R.
Equivalent diagnostic validity of ultrasound and dual x-ray
absorptiometry in a clinical case-comparison study of women
with vertebral osteoporosis. J Bone Miner Res 1994;9(Suppl
l):S211.
62. Gonnelli S, Cepollaro C, Agnusdei D, Palmieri R, Rossi S,
Gennari C. Diagnostic value of ultrasound analysis and bone
densitometry as predictors of vertebral deformity in postmeno-
pausal women. Osteoporosis Int 1995;5:413-8.
63. Stegman MR, Heaney RP, Recker RR. Comparison of speed of
sound ultrasound with single photon absorptiometry for deter-
mining odds ratio. J Bone Miner Res 1995;10:346-52.
64. Heaney RP, Avioti LV, Chesnut CH, Lappe J, Recker RR,
Brandenburger GH. Ultrasound velocity through bone predicts
incident vertebral deformity. J Bone Miner Res 1995;10:341-5.
65. Stegman MR, Heaney RP, Travers-Gustafson D, Leist J.
Cortical ultrasound velocity as an indicator of bone status.
Osteoporosis Int 1995;5:349-53.
66. Orgee J, McCloskey EV, Foster H, Coombes G, Khan S, Kanis
JA. Tibial ultrasound velocity: a useful clinical measure of
skeletal status. J Bone Miner Res 1994;9:S156.
67. Hans D, Dargent P, Schott AM, et al. Ultrasound parameters
predict hip fracture independently of hip bone density: the
EPIDOS prospective study. J Bone Miner Res 1995; 10(Suppl):
$169.
Ultrasound in the Assessment of Osteoporosis
68. Garcia B J, Foster FS, McNeill RG. Ultrasonics attenuation in
bone. Ultrasonics symposium proceedings, 1978:327-30
69. Smith SW, Phillips D J, Von Ramm OT, Thurstone FL. Some
advances in acoustic imaging through the skull: In: Linzer M,
editor. Ultrasonic tissue characterization II. Washington: Natio-
nal Bureau of Standards Special Publication 525:1979:209-217.
70. McCloskey EV, Murray SA, Charlesworth D, et al. Assessment
of broadband attenuation in the os calcis in vitro. Clin Sci
1990;78:221-5.
71. McKelvie ML, Fordham J, Clifford C, Palmer SW. In vitro
comparison of quantitative computer tomography and broad-
band ultrasonic attenuation of trabecular bone. Bone
1989;10:101-4.
72. Langton CM, Njeh CF, Hodgskinson R, Currey JD. Prediction
of mechanical properties of the human calcaneus by broadband
ultrasonic attenuation. Bone 1996;18:495-503.
73. Bouxsein ML, Courtney AC, Hayes WC. Ultrasound and
densitometry of the calcaneus correlate with the failure loads of
cadaveric femurs. Calcif Tissue Int 1995;56:99-103.
74. Ross P, Huang C, Davis J, et al. Predicting vertebral deformity
using bone densitometry at various skeletal sites and calcaneus
ultrasound. Bone 1995;16:325-32.
75. Kleerekoper M, Nelson DA, Flynn MJ, Pawluszka AS, Jacob-
sen G, Peterson EL. Comparison of radiographic absorptio-
metry with dual-energy X-ray absorptiometry and quantitative
computed tomography in normal older white and black women.
J Bone Miner Res 1994;9:1745-9.
76. Gluer CC, Vahlensieck M, Faulkner KG, Engelke K, Black D,
Genant HK. Site-matched calcaneal measurement of broadband
ultrasound attenuation and single X-ray absorptiometry: do they
measure different skeletal properties? J Bone Miner Res
1992;7:1071-9.
77. Brandenburger GH. Clinical determination of bone quality: is
ultrasound an answer? Calcf Tissue Int 1993;53(Suppl 1):$151-
6.
78. Faulkner KG, McClung MR, Coleman LJ, Kingston-Sandahl E.
Quantitative ultrasound of the heel: Correlation with densito-
metric measurements at different skeletal sites. Osteoporosis Int
1994;4:42-7.
79. Armitage P, Berry G. Statistical methods in medical research.
Oxford: Blackwell Scientific Publications, 1987.
80. Langton CM. The role of Ultrasound in the assessment of
osteoporosis. Clin Rheumatol, 1994;13(Suppl 1):13-7.
81. Resch H, Pietscbmann P, Bernecker P, Krexner E, Wilvonseda
R. Broadband ultrasound attenuation: a new diagnostic method
in osteoporosis. AJR 1990;155:825-8.
82. Herd RJM, Ramalingham T, Ryan PJ, Fogelman I, Blake GM.
Measurements of BUA in the calcaneus in premenopausal and
postmenopansal women. Osteoporosis Int 1992;2:247-51.
83. Damilakis JE, Drekatis E, Courtsoyiannis NC. Ultrasound
attenuation of the calcaneus in female population: normative
data. Calcif Tissue Int 1992;51:180-3.
84. Baran DT, Kelly AM, Karellas A, et al. Ultrasonic attenuation
of the os calcis in women with osteoporosis and hip fractures.
Calcif Tissue Int 1988;43:138-42
85. McCloskey EV, Murray SA, Miller C, et al. Broadband
ultrasound attenuation in the os calcis: relationship to bone
mineral at other skeletal sites. Clin Sci 1990;78:227-33.
86. Agren M, Karellas A, Leahey D, Marks S, Baran DT.
Ultrasound attenuation of the calcaneus: a sensitive and specific
discriminator of osteopenia in postmenopausal women. Calcif
Tissue Int 1991;48:2404.
87. Funke M, Kopka L, Vosshenrich R, Fischer U, Ueberschaer A,
Oestmann JW, Grabbe E. Broadband ultrasound attenuation in
the diagnosis of osteoporosis: correlation with osteodensito-
metry and fracture. Radiology 1995;194:77-81.
88. Stewart A, Reid DM, Porter RW. Broadband ultrasound
attenuation and dual energy X-ray absorptiometry in patients
with hip fractures: which technique discriminates fracture risk?
Calcif Tissue Int 1994;54:466-9.
89. Orwoll ES, Oviatt SK, Mann T. The impact of osteophytic and
vascular calcification on vertebral mineral density measurement
in men. J Clin Endocrinol Metab 1990;70:1202-7.
21
90. Stewart A, Felsenberg, D, Kalidis L, Reid DM. Vertebral
fractures in men and women: how discriminative are bone mass
measurements? Br J Radiol 1995;68:614--20.
91. Bauer DC, Gluer CC, Genant HK, Stone K. Quantitative
ultrasound and vertebral fracture in postmenopausal women. J
Bone Miner Res 1995;10:353-8.
92. Dretakis EC, Kontakis GM, Steriopoutos CA, Dretakis CE.
Decreased broadband ultrasound attenuation of the calcaneus in
women with fragility fracture. Acta Orthop Scand 1994;65:305-
8.
93. Kroger H, Jurvelin J, Amala I, et al. Ultrasound attenuation of
the calcaneus in normal subjects and in patients with wrist
fracture. Acta Orthop Scand 1995;66:47-52.
94. Porter RW, Miller CG, Grainger D, Palmer SB. Prediction of
hip fracture in elderly women: a prospective study. BIVIJ
1990;301:638-41.
95. Bauer DC, Gluer CC, Pressman AR, et al. Broadband ultraso-
nic attenuation (BUA) and the risk of fracture: a prospective
study. J Bone Miner Res 1995;10(Suppl 1):$175.
96. Gluer CC, Fuerst T, Wu CY, et al. Diagnostic sensitivity of
various quantitative ultrasound and dual x-ray absorptiometry
approaches. J Bone Miner Res 1995, 10(Suppl 1):$373.
97. Fordham JN, Langton CM, Tulsidas H. Ultrasonic measure-
ments of bone density in rheumatoid arthritis and osteoarthritis.
In: Palmer SB, Langton CM, editors. Ultrasonic studies of bone.
Bristol: IOP Publication, 1987:47-53.
98. Nijs J, Boonen S, Geusens P, Borghs H, Dequeker J.
Ultrasound in osteoarthritis and osteoporosis. In: Ring EF,
Elvins DM, Bhalla AK, editors. Current research in osteopor-
osis and bone mineral measurement III. Bath: British Institute
of Radiology, 1994:45.
99. Acotto CG, Schott AM, Hans D, Njepomniszeze H, Mautalen
CA, Meunier PJ. Hyperthyroidism influences ultrasound bone
measurements of the calcaneus. J Bone Miner Res 1995;
10(Suppl 1):$400.
100. Jones PRM, Hardman AE, Hudson A, Norgan NG. Influence of
brisk walking on the broadband ultrasonic attenuation of the
calcaneus in previously sedentary women aged 30--61 years.
Calcif Tissue Int 1991 ;49:112-5.
101. Herd RJM, Blake GM, Ramalingham T, Miller CG, Ryan PJ,
Fogelman I. Measurements of postmenopausal bone loss with a
new contact ultrasound system. Calcif Tissue Int 1993;53:153-7.
102. Rosenthall L, Tenenhouse A, Caminis J. A correlative study of
ultrasound calcaneal and dual-energy x-ray absorptiometry bone
measurements of the lumbar spine and femur in 100 women. Eur
J Nucl Med 1995;22:402-6.
103. Langton CM, Evans GP, Hodgskinson R Riggs CM. Ultrasonic,
elastic and structural properties of cancellous bone. In: Ring
EFG, editor. Current research in osteoporosis and bone mineral
measurement. Bath: British Institute of Radiology, 1990.
104. Nicholson PHF, Haddaway M J, Davie MW. The dependence of
ultrasonic properties on orientation in human vertebral bone.
Phys Med Biol 1994;39:1013-24
105. Tavakoli MB, Evans JA. The effect of bone structure on
ultrasonic attenuation and velocity. Ultrasonics 1992;30:389-95.
106. Hans D, Aflot ME, Shott AM, Roux JP, Kotzki PO, Meunier
PJ. Do ultrasound measurements on the os calcis reflect more
bone microarcitecture than the bone mass? A two-dimensional
histomorphometric study. Bone 1995;16:295-300.
107. Gluer CC, Wu CY, Jergas M, Goldstein SA, Genant HK. Three
quantitative ultrasound parameters reflect bone structure. Calcif
Tissue Int 1994;55:46-52.
108. Laugier P, Giat P, Berger G. New ultrasonic methods of
quantitative assessment of bone status. Eur J Ultrasound
1994;1:23-38.
109. Petley GW, Robins PA, Aindow JD. Broadband ultrasonic
attenuation: are current measurement techniques inherently
inaccurate? Br J Radiol 1995;68:1212-4.
110. Laugier P, Giat P, Berger G. Broadband ultrasonic attenuation
imaging: a new imaging technique of the os calcis. Calcif Tissue
Int 1994;54:836.
111. Roux C, Fournier B, Langier P, et al. Ultrasound bone imaging:
clinical evaluation of skeletal status. Osteoporosis Int 1996;6:84.
22
112. Langton CM. Electronic phantom for the ultrasonic assessment
of bone. Osteoporosis Int 1996;6:87.
113. Njeh CF, Langton CM. Prediction of bone strength from
ultrasound velocity and apparent density. Osteoporosis Int
1996;6:83.
114. Hans D, Dargent P, Schott AM et al. Which ultrasound
parameters could be combined with DXA to provide a better
prediction of hip fracture in the elderly? The EPIDOS prospec-
tive study. Osteoporosis Int 1996;6:84.
115. Gnudi S, Malavolta N, Ripamonti C, Caudarella R. Ultrasound
in the evaluation of osteoporosis: a comparison with bone
mineral density at distal radius. Br J Radiol 1995;68:476-80.
116. Wapniarz M, Lehmann R, Banik N, Radwan M, Klein K,
Allolio B. Apparent velocity of ultrasound (AVU) at the patella
in comparison to bone mineral density at the lumbar spine in
normal males and females. Bone Miner 1993;23:243-52.
117. Fujii Y, Goto B,Takahashi K, Fujita T. Ultrasound transmission
as a sensitive indicator of bone change in Japanese women in the
perimenopausal period. Bone Miner 1994;25:93-101.
118, Foldes AJ, Rimon A, Keinan DD, Popovtzer MM. Quanitative
ultrasound of the tibia: a novel approach for assessment of bone
status. Bone 1995;17:363-7.
119. Herd RJM, Blake GM, Miller CG, Parker JC, Fogelman I. The
ultrasonic assessment of osteopenia as defined by dual x-ray
absorptiometry. Br J Radiol 1994;67:631-5.
120. Schott AM, Hans D, Sornay-Rendu E, Delmas PD, Meunier PJ.
Ultrasound measurements on os calcis: precision and age-
related changes in a normal female population. Osteoporosis Int
1993;3:249-54.
121. Salamone LM, Krall EA, Harris S, Dawson-Hughes B. Com-
parison of broadband ultrasound attenuation to single x-ray
C. F. Njeh et al.
absorptiometry measurements at the calcaneus in postmeno-
pausal women. Calcif Tissue Int 1994;54:87-90.
122. Evans WD, Crawley EO, Compston JE, Evans C, Owen GM.
Broadband ultrasonic attenuation and bone mineral density
(letter). Clin Phys Meas 1988;9:163-5
123. Petley GW, Hames TK, Cooper C Langton CM, Cawley MD. A
comparison of single photon absorptiometry and broadband
ultrasonic attenuation: past, present and future. In: Palmer SB,
Langton CM, editors. Ultrasonics studies of bone. Institute of
Physics Short Meetings no. 6, 1987.
124. Hosie CJ, Smith DA, Deacon AD, Langton CM. Comparison of
broadband ultrasonic attenuation of the os ealcis and quantita-
tive computed tomograpby of the distal radius. Calcif Tissue Int
1987;48:303-8.
125. Poll V, Cooper C, Cawley MID. Broadband ultrasonic attenu-
ation in the os calcis and single photon absorptiometry in the
distal forearm: a comparative study. Clin Phys Physiol Meas
1986;7:375-9.
126. Massie A, Reid DM, Porter RW. Screening for osteoporosis:
comparison between dual-energy X-ray absorptiomet~y and
broadband ultrasound atenuation in 100 perimenopausal
women. Osteoporosis Int 1993;3:107-10.
127. Young A, Howey S, Purdie DW. Broadband ultrasound attenu-
ation compared with dual-energy X-ray absorptiometry in
screening for postmenopausal low bone density. Osteoporosis
Int 1993;3:160-4.
128. Baran DT, McCarthy CK, Leahey D, Lew R. Broadband
ultrasound attenuation of the calcaneus predicts lumbar and
femoral neck density in caucasian women: a preliminary study.
Osteoporosis Int 1991;1:110-3.
Received for publication 10 January 1996
Accepted in revised form 20 August 1996