Ultrasonic attenuation in human tissue

Ultrasonic attenuation in human tissue

Ultrasound in Med. & Biol., Vol. 2, pp. 25-29. Pergamon Press, 1975. Printed in Great Britain. ULTRASONIC ATTENUATION IN HUMAN TISSUE R. C. CmvEgs* a...

571KB Sizes 3 Downloads 89 Views

Ultrasound in Med. & Biol., Vol. 2, pp. 25-29. Pergamon Press, 1975. Printed in Great Britain.

ULTRASONIC ATTENUATION IN HUMAN TISSUE R. C. CmvEgs* and C. R. HILL Physics Division, Institute of Cancer Research, Sutton, Surrey, England

(First received 10 December 1974; and in final form 12 March 1975)

Abstract--A spectroscopic method of measuring attenuation is described and measurements of attenuation as a continuous function of frequency between 1-0 and 7-5 MHz presented for fixed human fat, liver and spleen at 18°C. These are compared with data from the literature, showing approximate agreement for liver, whilst for fat the values are significantly higher than those previously reported.

Key words: Absorption, Acoustic, Attenuation, Fat, Liver, Scattering, Spectrum analysis, Spleen, Ultrasonics.

animals, this constituting the sum of information available to that time on liver attenuation in the frequency range of diagnostic interest (1-10 MHz). It is only recently (Pauly and Schwan, 1971 and Mountford and Wells, 1972) that any attempt has been made to confirm or extend these earlier results. The majority of the measurements reported have been collated by Wells (1975). In the second place the data is of little help to later workers since few reports give details of the preparation of the specimens and some omit important parameters. The most frequent truant is the temperature at which the measurements were made. Dunn et al. quote a factor of seven for the change in attenuation co-efficient of tissues between 0 ° and 45°C. The use of measurements made at room temperature will not necessarily be commensurate with those made in vivo (e.g. Mountford and Wells, 1972). In the third place, and most important of all, the authors describing methods of measurements tend to claim to measure attenuation whereas the results reported are all termed absorption measurements. As has been discussed above, these two terms are not necessarily consistent. The ultrasound that is scattered back from the internal structure of a volume of one type of soft tissue is measurable and has been considered by several authors (Hill and Chivers, 1972; Chivers et al., 1973; Mountford and Wells, 1972a, b) as a potential source of diagnostic information.


The propagation of a beam of ultrasound in a biological medium may be attenuated by 3 main factors: geometry, scatter and absorption. Included in the first are the properties of the radiation field of the transducer used in making the measurements (e.g. beam divergence and diffraction) and any reflection and refraction occurring at macroscopic boundaries of the medium (i.e. those of dimension of the order or larger than a wavelength). In any careful measurements of attenuation these will be minimised and corrections applied to remove their effects. The corrected measurements will thus give the attenuation caused by the scattering of ultrasound out of the main beam by discontinuities of size comparable with, or smaller than the wavelength, and by the absorption of energy by the tissue. This latter will involve a conversion of the energy from ultrasonic vibrations into molecular vibration and heat. According to Dunn et aL (1969), the literature is "replete with data regarding the dependence of the ultrasound absorption coefficient upon the acoustic frequency". It would appear, however, that the data is not as complete as may be desired. In the first place the data is sparse. Wells' (1969) graphical representation of the tabular data of Goldman and Hueter (1956) shows eight measurements on livers from various * Present address: Physics Department, University of Surrey, Guiidford, Surrey, U.K. 25

\ U.M.B. Vol. 2 N o . I~C


R . C . CHIVERS and C. R. HILL

The need for a systematic set of measurements of ultrasonic attenuation of human tissue is indicated by present developments in the field of diagnosis. The existing data has been adequate for order of magnitude calculations for time-varied gain which was subsequently optimised by experience. In addition to being a pre-requisite for fundamental work on the acoustical structure of tissue (Chivers et al., 1973), attenuation data is becoming of increasing significance in the sophisticated techniques being developed. The use of grey-scale display in B-scanning (Kossoff, 1974) requires information on the range of values that the attenuation coefficient may take in clinical situations, while the increasing use of a computer for the storage and display of anatomical cross-sections (Milan, 1972) enables more accurate correction factors to be applied if the basic data is available. As a source of data for testing hypotheses concerning absorption mechanisms (Fry, 1952; Pauly and Schwan, 1971) the literature has not been helpful, perhaps because of the scarcity of data and the lack Of numerical information on tissue scattering. As the latter becomes available, this very important area can be investigated more thoroughly than has been possible up to the present time. The field of dosimetry will also benefit from these measurements, since the parameter of interest is the absorbed dose, requiring values of the absorption coefficient (rather than the attenuation coefficient usually used) for multiplication with the incident intensity. METHOD

The equipment used is illustrated in Fig. 1. A pulse of ultrasound emitted by a transducer, taken from a diagnostic machine in routine use, was reflected from a plane target, and the reflected pulse isolated by a time-gating circuit prior to feeding it into a spectrum analyser. The analyser displays the Fourier transform of the pulse reflected from the plane target. The attenuation of a parallel-sided tissue specimen is the difference (if a logarithmic spectral display is used) between the target echo with and without the sample interposed between the transducer and the target. The duration of the pulse returned from the target was always sufficiently short both to exclude the possibility of significant standing wave effects and to lie completely within the gate length used (8 rtsec), so that its spectrum was virtually unaffected by the existence of the gate. This system (Chivers, 1973) is similar to that developed independently by




: Itransi iducer i~


i Trigger~ t



tissue reflector I ~:~L- ~ / [Ji i ~, = Sound Ii(~:,i:i ~ !' Tank




:i j

A Gate ~ Speetn~m Analyser


] Frequency

Fig. I. A diagrammatic representation of the experimental system.

Papadakis et al. (1973) for measuring attenuation in metals. Movement of the target within the near field of the transducer showed no variation of the spectrum of the echo it produced (to within approx. 0.5 dB) over the range of frequencies of the transducer bandwidth. A list of the diameters, nominal frequencies and usable bandwidths of the probes (as just defined) is given in Table 1. Within these specified bandwidths, measurements can be made as a continuous function of frequency. The overlap of bandwidths and consistency of results obtained helps remove suspicion of geometrical factors significantly affecting the results obtained. The spectrum of the echo from the plain Perspex target for the 2 MHz transducer is shown in Fig. 2. Measurements were made at 18 + 2°C in a tank of degassed water (to discourage bubble formation). The samples were post mortem specimens that had been fixed in formalin for several weeks prior to use. Two samples of each type of tissue were cut and suspended in a water filled polythene bag supported by a Perspex frame, The width of the frame was adjusted to the thickness of the specimen so that the bag was in close but not oppressive contact with the Table !. Nominal frequency, diameter and bandwidth of transducers Nominal frequency (MI-l.z)

Dia. (mm)

Effective bandwidth (MHz)

% bandwidth

1.0 1.5 2"0 4-0 6.0

20 20 15 15 10

0.6-1.4 1.0-2.5 1'6-2'8 3"0-6'0 4.0-7.5

40 43 27 33 30

Ultrasonic attenuation in h u m a n tissue


Log. Scale lOdB/div.




3.0 M H z


I Usable Bandwidth

Fig. 2. The spectrum of the echo from a plane target for a transducer of nominal frequency 2 MHz, taken with a 30 k H z filter

tissue on each side. At least three measurements were made in different regions on each sample so that, in the regions where the transducer bandwidths overlapped, the mean has been taken over at least 12 readings. There was no significant difference between measurements made on different samples of the same type of tissue. RESULTS

The curves obtained of attenuation (measured in dB/cm) for fixed human fat, liver and spleen as a continuous function of frequency are shown in Fig. 3. The error bars give the standard

deviation of the readings taken and have been included to show the order of magnitude of the variation observed. The continuous curves were drawn by eye. The tendency for both the fat and spleen attenuation to curve upwards suggests that the traditional assumption of the linearity of attenuation with frequency may only be an approximation. Certainly, more data would be required before a definitive statement can be made. Errors may arise from several sources, Geometrical errors and those associated with enlarging, tracing and reading the spectra are all encompassed in the approx. 0.5 dB quoted for the uniformity of the echo from the plane target

Liver I


Fat Spleen


// /jr.


] / .

// ./'/"

0 0

~.::: "'"~"







3 4 5 Frequency ( MHz )



Fig. 3. The attenuation of fixed h u m a n fat, liver and spleen as a continuous function o f frequency in the range 1.0-7.5 MHz.




R.C. CmVEaS and C. R. HILL

since this figure represents the spread of readings measured after these processes had been completed. It does not, however, include the small loss that occurs by reflection at the water: polythene: tissue interfaces which may be estimated to total 0.1 dB. The most serious problem arises from the specimen itself. Its orientation was adjusted to give a maximum echo from the front surface, and slight errors here involve a factor of cos 0 in the thickness--a fairly insensitive quantity when 0 is close to zero. It is almost certainly swamped by the error in the measurement of the thickness of the specimen. The flatness and parallel nature of the surfaces of the sample was as good as the eye and hand could provide, the cutting being effected with a microtone blade. Thickness measurements were taken both with callipers and with a depth gauge (for the latter the specimen was lying on a horizontal flat surface). Non-uniformity of thickness may have reached 1 mm in extent, and on specimens ranging from 1 to 3 cm in thickness, represented up to 10 per cent error. The curves drawn in figure three are replott~ on Fig. 4, which shows the summary of Goldman and Hueter (1956), together with points from Pauly and Schwan (1971) for beef liver homogenate. The general tendency for the fixed tissue points to lie above those for fresh tissue may be indicative of structural changes taking place in the fixing process, but a close comparison of the results is not necessarily


valid in view of the different circumstances under which the experiments may have been done, as mentioned above. Experiments at much higher frequencies (Kessler, 1973) suggest that the condition of the tissue, i.e. fixed, fresh or thawed) does not significantly affect the results of the measurements. This question is under investigation for the frequencies of interest in diagnosis. DISCUSSION The technique described here has a number of useful features, in that it has potentially greater accuracy than the long pulse technique of Dunn et al. (1969) and can monitor a large range of frequencies in a continuous manner. If a specimen 3 cm long with accurately parallel sides could be obtained, the accuracy of measurement could reach 0.04 dB/cm (compared to 0" 15-0.5 dB/cm for the data reported here). Use of a linear spectral display might provide further improvement in accuracy at the cost of reducing the usable bandwidth of each transducer and extending the calculations involved. Results obtained using this technique would be comparable to those obtained using narrow bandwidth techniques; they will not, however, be commensurate with those made using the decrement of amplitude of a short pulse as a measure of attenuation (e.g. Mountford and Wells, 1972). The short pulse used will contain a broad range of frequencies (see Fig. 2 and Table 1) and its shape will inevitably change as it proceeds through a medium that provides dispersive attenuation such as tissue. The attenuation measured by such a technique will thus depend on the shape of the pulse input. For a diagnostic situation, it will be clearly meaningful, but may vary from individual machine to individual machine. The virtue of the narrow band techniques such as the one described here is the fundamental information they provide.


Acknowledgements--One of the authors (R.C.C.) would like to thank the Royal Marsden Hospital for financial support for this work which was carried out in the Physics Division of the Instituteof Cancer Research,Sutton, Surrey, with the support and encouragement of Professor J. W.

........ ...'"





1.0 Frequency (MHz)

I I I I llll




F r e s h and fixed tissue (Goldman and Hueter, [email protected]) Fresh beef liver {Pauly and Schwan, 1971) Fixed human liver "[

........... ............

Fixed human fat / Data l r o m Fixed human spleen Fig. 3

Fig. 4. Comparison of experimental data on attenuation with that in the literature.


Chivers, R. C. (1973) The scattering of ultrasound by human tissues, Ph.D. Thesis, Universityof London. Chivers,R. C., Hill,C. R. and Nicholas,D. (1973) Frequency dependence of ultrasonic backscatteringcross-section:an indicator of tissue structure characteristics. In Ultrasonics in Medicine (Edited by de Vlieger, M. et al.), pp. 300-303. Excerpta Medica,Amsterdam.

Ultrasonic attenuation in human tissue Dunn, F., Edmonds, P. D. and Fry. W. J. (1969) In Biological Engineering (Edited by Schwan, H. P.), pp. 205-332. McGraw-Hill, New York. Fry, W. J. (1952) Mechanism of acoustic absorption in tissue. J. acoust. Soc. Am. 24. 412-415. Goldman, D. E. and Hueter, T. F. (1956) Tabular data of the velocity and absorption of high frequency sound in mammalian tissues. J. acoust. Soc. Am. 28,.35-37. Hill, C. R. and Chivers, R. C. (1972) In Ultrasonics in Biology and Medicine (Edited by Filipczynski, L.), p. 120. Polish Scientific Publishers, Warsaw. Kessler, L. W. (1973) Very high frequency ultrasonic attenuation in mammalian tissue. J. acoust. Soc. A m 53, 1759-1760. Kossoff, G. (1974) Ultrasonic visualization of the uterus, breast and eye by grey scale echography. Proc. R. Soc. Med. 67, 135-140. Milan, J. (1972) An improved ultrasonic scanning system


employing a small computer. Br. J. Radiol. 45, 911-916. Mountford, R. A. and Wells, P. N. T. (1972a) Ultrasonic liver scanning: the quantitative analysis of the normal A-scan. Phys. Med. Biol. 17, 14-25; (1972b) Ultrasonic liver scanning: the A-scan in the normal and cirrhosis. Phys. Med. Biol. 17, 261-269. Papadakis, E. P., Fowler, K. A. and Lynworth, L. C. (1973) Ultrasonic attenuation by spectrum analysis of pulses in buffer rods: method and diffraction corrections. J. acoust. Soc. Am. 53, 1336-1343. Pauly, H. and Schwan, H. P. (1971) Mechanism of absorption of ultrasound in liver tissue. J. acoust. Soc. Am. 50, 692-699. Wells, P. N. T. (1969) Physical Principles of Ultrasonic Diagnosis. Academic Press. New York. Wells. P. N. T. (1975) Absorption and dispersion of ultrasound in biological tissue. Ultrasound Med. Biol. 1, 369-376.