Page 40 - Summer 2018

Quantitative Ultrasound and Osteoporosis
Figure 6. Ultrasound signals from through-transmission measurements of cancellous bone sample for thicknesses rang- ing from 0.5 to 11.8 mm. Signals were obtained by alternately performing through-transmission measurements and shaving approximately 0.5 mm from the bone sample. The normal- ized amplitude is magnified to show both fast (low-amplitude; left) and slow (high-amplitude; right) waves. For the thickest samples, the fast and slow waves are well separated. For sam- ples thinner than 6 mm, however, fast and slow waves begin to overlap (Groopman et al., 2015).
has been demonstrated in cancellous bone (Williams, 1992; Hosokawa and Otani, 1997; Lee et al., 2003; Figure 6). Be- cause fast and slow waves often overlap in both the time and frequency domains, advanced signal-processing methods have been developed to separate them (Groopman et al., 2015; Figure 7).
The Interaction of Ultrasound
with Cortical Bone
Calcaneal devices, which target cancellous bone, were in- troduced about 20 years ago and have since received formal recognition of their clinical effectiveness from professional organizations (Krieg et al., 2008; US Preventive Services Task Force, 2011). Many devices that target cortical bone are newer and still undergoing research and development. Cor- tical devices aim to measure changes in cortical thickness and porosity, both of which have been shown to be related to fracture risk.
Because it is a projection technique, DXA has limited ca- pability to provide reliable quantitative measurements on cortical bone. High-resolution (HR) pQCT (HR-pQCT)
38 | Acoustics Today | Summer 2018
Figure 7. Fast- and slow- wave decomposition of the data from Figure 6. Phase velocity (top) and nBUA (bottom) of fast (left) and slow (right) waves are functions of sample thickness. Bayesian and Prony’s (modified least-squares Prony’s plus curve-fitting \\\\\\\[MLSP+CF\\\\\\\]) methods can recover wave proper- ties at thicknesses below 6 mm even when fast and slow waves overlap and conventional methods fail. Bayesian and Prony’s methods show high agreement with each other and provide plausible extrapolations of properties for thicknesses below 6 mm (Groopman et al., 2015).
provides full three-dimensional information but is expen- sive and limited to clinical research facilities.
Ultrasound waves transmitted by a source through the skin to a long bone (such as the radius or the tibia) can gener- ate vibrations that propagate in the cortex along the axis of the bone. As they propagate, these guided waves leak energy from the waveguide to the adjacent soft tissue. The leaked energy can be detected using sensors placed on the skin, typically a few centimeters away from the source. So-called axial transmission (AT) methods have been developed to measure these guided waves.
An early approach used point contact transducers with ex- ponential waveguides to measure the speeds of surface and flexural waves at the mid tibia at 100 kHz. Guided waves were used to measure patients with immobilization atrophy (Dzene et al., 1980) and to monitor skeletal demineralization of cos- monautsexposedtomicrogravity(Tatarinovetal.,1990).