| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Rheumatology 2001; 40: 829-831
© 2001 British Society for Rheumatology
Letters to the Editor |
Comment on ‘ultrasonic measurement of the thickness of human articular cartilage in situ’ by Yao and Seedhom
Mechanical Engineering, Room 3-137D, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA
SIR, In their abstract the authors say under Methods ‘The velocity of sound in articular cartilage was measured ...’, then under Results ‘the velocity of sound ... varied widely’, and in their Conclusion ‘The ultrasonic-pulse-echo is not accurate for the measurement of the thickness of cartilage in situ’ [1]. In fact, the authors never measured the velocity of sound in cartilage directly (my italics). Rather, the authors defined ‘A mesh ... drawn on the articular surface using a soft, felt-tipped pen’. Then, using a commercial ultrasonic device, ‘the transducer was pointed [manually?] perpendicularly to the articular surface’ at each mesh location. From the echoes from the cartilage surface and the cartilage–bone interface (no wave forms were shown), ‘the time period for the sound to travel through the thickness of the cartilage layer was obtained’. The cartilage layer thickness at each mesh location was measured using their (preferred) needle penetration method. They then used the needle measurement and the prior ultrasonically based time interval to calculate the velocity of sound at each location.
The ultrasonic thickness measurement technique employed by Yao and Seedhom was seriously flawed in at least four ways. First, and fundamentally, their measurement of the velocity of sound in the cartilage layer was very indirect. By comparison, Rushfeldt in his 1978 MIT ScD thesis [2] (which we believe to describe the first use of ultrasound to measure cartilage thickness in situ) and as reported in a subsequent publication [3], determined the velocity of sound directly by measuring the sound wave transit time though measured thicknesses of cartilage, before applying the technique to acetabular cartilage in situ. Thirty cartilage samples were machined in a microtome to form parallel-sided discs and the sound velocity was determined to be 1759 m/s with a standard deviation of 90 m/s, less than 4% of the average value. Modest, in her MIT bachelor's thesis [4] and a subsequent publication [5], verified the accuracy of the ultrasonic technique by performing direct comparisons of sound vs optical measurements on cartilage–bone cores, in part to confirm that the ‘second’ maximum of the ultrasonic waveform was the reflection from the calcified cartilage layer. To quote from reference 5, ‘For all but three of the [24] samples, the ultrasound measurement falls well within one standard deviation of the mean plug thickness based on the optical measurements at the perimeter (of the core)’, (see Fig. 5 in [5]).
Second, it is difficult to establish the perpendicularity of the transducer axis to the cartilage surface; any error changes (increases) the sound path length through the cartilage layer, and therefore the sound transit time. In his study of acetabular cartilage thickness, thickness distribution and cavity topology (sphericity), Rushfeldt mounted the transducer in a custom-built hip simulator [6], which ensured that the transducer moved in a spherical coordinate system relative to the acetabulum. Tepic et al. [7] adapted the Rushfeldt ultrasound system to measure cartilage layer modulus and permeability in situ. This technique made even greater demands on perpendicularity, because the amplitude of the reflected signal was used to determine the mechanical properties of the cartilage layer. Tepic designed a unique conical-scan ultrasonic transducer to ensure that waveform data were recorded when the sound beam was perpendicular to the cartilage surface.
Third, the locational correspondence of each of the Yao and Seedhom ultrasonic and penetration trials was inadequately controlled, based on ‘A mesh drawn ... with a felt-tipped pen’. The radical difference between how the two methods detect the cartilage–bone interface further challenges the validity of using the needle penetration measurement as the basis for translating the ultrasonic time data into sound velocity. The sound waveform represents energy reflected from a considerable area of the cartilage–bone or calcified cartilage interface, larger by orders of magnitude than the area of the cartilage–bone interface sensed by the sharp-tipped needle. Whereas the cartilage surface is relatively smooth, the cartilage–bone interface is very irregular on a small scale. Thus the sonic data average the cartilage thickness over a large area while the needle may contact the bone at a small prominence or enter a deep cavity, introducing a large amount of variance into the needle measurements. Modest's optical measurements around the perimeter of each 5 mm core depicted this variation, typically more than 1 mm in a total cartilage layer thickness of 1.6 mm (Fig. 3 in [5]).
Fourth, the description of the ultrasonic system employed by Yao and Seedhom is limited to the name and model of the commercial ultrasonic transducer they used. Rushfeldt [2, 3, 6], Tepic [7–10], Macirowski [11, 12] and Modest [4, 5] at MIT also used a commercial transducer and pulser/receiver. However, they integrated these with extensive commercial and custom components and software in a system which fully automated data acquisition and reduction. Tepic et al. [8–10] and Macirowski et al. [11, 12] in their integrated system for measuring the geometry and mechanical properties of cartilage layers in situ sampled and digitized all waveform data at 100 MHz with a Biomation Waveform Recorder (Gould, East Lake, OH, USA), improving the signal-to-noise ratio by ensemble averaging of 64 or 128 waveform records at each location, followed by computer storage. Analysis of the signal was performed off-line using a correlated receiver technique [13]. A standard reflection signature for the ultrasonic transducer was recorded from a large, precise metal target (also used for ultrasonic error determination). For each location on the hip joint cartilage (about 300 on the femoral head and 100 on the acetabulum) the signature signal was cross-correlated with the stored waveform, using Fast Fourier Transforms to identify the maxima corresponding to the cartilage surface and calcified interface for geometry, and normalized amplitude for the modulus and permeability determination. Whittaker's reconstruction [14] was used to interpolate between sample points, increasing the effective resolution of the system. Figure 1 of reference 5 is a schematic of the ultrasonic measurement system. These methods improved the resolution of the MIT ultrasound system to 1.5 µm.
Thus, when one considers how Yao and Seedhom calculated indirectly the sound velocity in cartilage, using time intervals produced by a commercial device (undoubtedly intended for some other ultrasound application), the problem of ensuring consistent perpendicularity of transducer axis and cartilage surface, the gross (felt pen) matrix for locating the supposedly corresponding sound and needle measurements and the inescapable disparity between how sound reflection vs needle-point define the cartilage–bone interface, the huge ‘error in the thickness of cartilage obtained from the ultrasonic method’ their report comes as no surprise, and is clearly a result of their flawed methodology, as described in their Rheumatology article and in their prior abstract in Connective Tissue Research [15].
Much the same criticisms of the ultrasonic measurement method described above for the Yao and Seedhom reports can be applied to the method and conclusions of Jurvelin et al. [16] (cited by Yao and Seedhom), who in their abstract also ‘invalidate the use of the A-mode, 10 MHz-ultrasound device for thickness measurements’. At least Jurvelin et al. cite the references 3 and 5 listed here; Yao and Seedhom cite none of the MIT references in either of their articles.
The MIT theses referenced include complete details of the methods, systems and software developed and applied, including computer codes. The theses are available from the MIT Library System.
References
-
Yao JQ, Seedhom BB. Ultrasonic measurement of the thickness of human articular cartilage in situ. Rheumatology1999;38:1269–71.
[Abstract/Free Full Text] - Rushfeldt PD. Human hip joint geometry and the resulting pressure distributions. ScD thesis, Department of Mechanical Engineering, Massachusetts Institute of Technology, 1978.
- Rushfeldt PD, Mann RW, Harris WH. Improved techniques for measuring in vitro the geometry and pressure distribution in the human acetabulum. I. Ultrasonic measurement of acetabular surfaces, sphericity and cartilage thickness. J Biomech1981; 24:253–60.
- Modest VE. Optical verification of computer-mediated ultrasonic thickness measurements of unloaded cartilage layers in situ. SB thesis, Department of Mechanical Engineering, Massachusetts Institute of Technology, 1986.
- Modest VE, Murphy MC, Mann RW. Optical verification of a technique for in situ ultrasonic measurements of articular cartilage thickness. J Biomech1989;22:171–6.[Medline]
- Rushfeldt PD, Palmer DW, Mann RW, Harris WH. In vitro hip joint pressure distribution testing facility. Proceedings of the Association for Advancement of Medical Instrumentation Conference, March 1978. Arlington, VA: Association for Advanced Medical Instrumentation, 1978:225.
-
Tepic S, Macirowski TJ, Mann RW. Articular cartilage mechanical properties elucidated by osmotic loading and ultrasound. Proc Natl Acad Sci USA1983;80:3331–3.
[Abstract/Free Full Text] - Tepic S. Congruency of the human hip joint. MSc thesis, Department of Mechanical Engineering, Massachusetts Institute of Technology, 1980.
- Tepic S. Dynamics of and entropy production in the cartilage layers of the synovial joints. ScD thesis, Department of Mechanical Engineering, Massachusetts Institute of Technology, 1982.
- Tepic S, Macirowski TJ, Mann RW. Computer simulation of interarticular fluid flow. Biomechanics: current interdisciplinary research. Hingham, MA: Kluwer, 1985:221–6.
- Macirowski TJ. Stress in the cartilage of the human hip joint. ScD thesis, Department of Mechanical Engineering, Massachusetts Institute of Technology, 1983.
- Macirowski TJ, Tepic S, Mann RW. Cartilage stresses in the human hip joint. J Biomech Eng1994;116:10–18.[ISI][Medline]
- Kak AC, Fry FJ, Jones JP. Acoustic impulse profiling. In: Fry FJ, ed. Ultrasound: its applications in medicine and biology, Chap. VIII. Amsterdam: Elsevier, 1978:495–537.
- Stearns SD. Digital signal analysis. Rochelle Park, NJ: Hayden, 1975.
- Yao JQ, Seedhom BB. Ultrasonic measurements of articular cartilage. Connect Tissue Res1997;36:143.
- Jurvelin JS, Rasansen T, Kolmonen P, Lyyra T. Comparison of optical, needle probe and ultrasonic techniques for the measurement of articular cartilage thickness. J Biomech1995;28:231–5.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||