JPS61247436A - Ultrasonic medium characteristic measuring device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

[Technical Field of the Invention] The present invention relates to an apparatus for measuring tissue characteristics of a living body, etc., and in particular, in response to the development of approximate formulas and theoretical formulas that describe the characteristics, characteristic parameters included therein can be expressed as spectral parameters. This invention relates to a method for optimizing and determining the number of channels and the number of channels. The 0th order is a particularly stable spectral parameter.
We will focus on the case where the first moment and the first moment are used. [Prior art to the invention] In order to obtain the ultrasonic characteristic values of an ultrasonic medium such as a living body,
An ultrasonic pulse is transmitted into the medium, and as the pulse progresses through the medium, the reflected waves are received from each depth and the medium characteristic values at each depth are measured. ing. In living organisms, non-invasive diagnosis of lesions is possible, and ultrasonic TC (Tissue Character
erization). Attenuation coefficients and reflection coefficients are the most commonly focused on ultrasonic characteristic values of media. The attenuation coefficient is △2 for ultrasonic sound pressure.
When transmitted through the thickness of , the transmitted sound pressure p is 1) - Po EXp (ct△z)
It is defined by α when given by (1). α is generally a function of frequency, but its functional form may be given theoretically, but most often it is given experimentally. In living organisms, α=βf (1) (
2) In liquids, α-ct f ” (1)
(2-1). However, as the frequency of the ultrasonic waves used expands to 1 to 10 MHz and measurement technology advances further, ex-cxf'' (m=1 to 2)
(2-2) has come to be considered to be in better agreement with experiments. Since this is an experimental formula, other formulas may be developed in the future, such as α=Af+
Br' (2) (2-3) α=βf+δ
(4) (2-4)ot=txe'f'
(4) (2-5) etc. may be used, and a theoretical formula may also appear. Equation (2) or (2-2) is generally used. The reflection coefficient is generally defined in terms of power, and the incident power is 1p.
ol" and the reflected power is IP-1", the power reflection coefficient, r, is defined as r=lP, l"/IP, l" (3). r is generally a function of frequency, and when scatterers are irregularly distributed close to each other, such as in a living body, the reflected waves from each scatterer overlap and interfere with each other, resulting in abnormal spectral irregularities - the so-called spectrum.・Scarosobu appears. This scallop can be reduced by averaging a large number of power spectra of reflected waves from nearby tissues around the region of interest. Regarding the power reflection coefficient after averaging like q, its frequency dependence has been determined theoretically and experimentally, and various formulas have been proposed, but there is a possibility that it will change further in the future. r==b −f' ・Exp (−(d f”))(
6) (4-1) = b-r' ・□ (6) (
4-2) 1・+ef” r-b (1+g f”) (n-3, or4)
(5) (4-3)=b (1+h f' + j
rq) (5) (4-4) Equation (4) is most widely used. In a case such as a living body, the path along which the ultrasound pulse travels, ie, the scan line (Z-axis), passes through various organs. In general, there is a smooth membrane at the boundary between organs, and the interface exhibits strong specular reflection (compared to the organ parenchyma) without frequency characteristics. ), it exhibits a fairly clear scallop characteristic due to mutual interference. However, the power reflectance is only a few percent, and most of the ultrasonic beams are transmitted. This transmittance during operation is τi (i is the i-th boundary counting from 2=0)
If τL has a small reflection loss, it can be approximated as having almost no frequency characteristics. Assuming that the transmittance when a reflected wave from a deeper part passes through this boundary in the opposite direction is τl, the same thing as τ5 can be said for τi. Attenuation coefficient (2) Equation 9 Reflection coefficient (4) Equation above. Using the spherical transmittance τ5, τi, the received echo, power spectrum IP,l'' is given as a function of z and frequency f by: I Poct+l'' is the power spectrum of the transmitted pulse to the medium surface It is. k (II) = (πτ positive τ5)b. The second term and subsequent terms on the right side of equation (5) are the power transfer functions of biological tissue, and are expressed as TTF (Tissue TransferF
It is called urnchction. Equation (5) is expressed as lPO(
This shows that ll+fl l 2 is observed. Below, we will describe a conventional method for determining medium characteristic values of ecological tissues from observations using such frequency windows. Initially, the attenuation coefficient is given by equation (2), and the reflection coefficient is (
In equation 4), α-β(3) ・f(2) r-k)(g)"'(l n"0+ or n=con
st (4) and n=0. That is, it has no frequency dependence. Miwa et al. proposed a method for determining β on the assumption that n is constant and does not change in all ranges of 2. This method provides windows with two different center frequencies f, f2. energy at,
That is, the 0th moment of the power spectrum (Mo)+,
(By finding Mo > t and taking the ratio of the two, (
5) Unknown quantities k (g) and r(,)(bn++
Or delete b(,,f'1) and get β3. This is a method to find . This can be processed in both the time domain and frequency domain. Furthermore, as research on biological characteristics using ultrasound progressed, the approximation of equation (4) became widely accepted, and as the data on the value of n became available, more general solutions were proposed. Hayakawa et al. [8] used three narrow-band windows (f+, fz, and fs are the respective frequencies) as the windows, and selected a spherical reflected wave that indicates the reflected intensity as the reflected wave, and calculated its bona fide power. We proposed to find the spatially averaged attenuation-related parameter d2 α d(lnf)'' of the parenchymal part between adjacent subicular reflection points from the ratio of . , this parameter becomes 2T, and when m=1, it becomes β in equation (2). However, n(
, ) has been proposed, but the distribution of the attenuation coefficient within the parenchyma has not been determined. Furthermore, because of the narrow band, errors are likely to occur due to the influence of scallops, and when extended to time domain processing, there is a drawback that the spatial resolution in the depth direction deteriorates. Miwa et al.

[9] has three broadband windows (each center frequency f
In fz, fa), create two equations by first eliminating k (g) from the ratio of their respective powers (energy key, 0th moment of power spectrum), and then calculate β, n using m=1 ．． J! We proposed the case of IJ α−βf. Compared to the method of Hayakawa [8], it rather eliminates subicular reflections, and for the parenchymal part, not only β but also n
can also be found. Because of the wide band, the effects of scalloping are weakened, and depth resolution is good even in time domain processing. This processing is possible in both the time domain and 9 frequency domains. However, Miwa et al.

The method [9] cannot be applied when α = zf'' (2-2). The present invention aims to provide a solution for this case. R. Kuc et al. [10] and Miwa et al. [11] uses one broadband Gaussian form when m = 1 and n = 0. or n = const.In addition to the 0th moment l-M, the first moment Ml is also extracted, and below the center frequency =
Find M+/Mo and calculate β3 from the change as a function of 2 below.
．． ) was proposed. Kuc (10) calculated wide spatial average values for large organs such as the liver, and Miwa et al. [11]
is looking for a local value. Furthermore, Nakajiro et al. [12] proposed a processing method in the time domain. However, it cannot be applied when n changes as a function of 2, when m≠1, etc. The present invention provides a solution for this case. Miwa et al. [13] used a one-channel broadband Gaussian form for the case m-1 and n = n. Extract the 0th, 1st, and 2nd (3rd) moments as spectral parameters, and further below the center frequency. Dispersion Σ2. We have proposed a method in which the skew S9, etc. are determined, and β and n are then determined as medium characteristic values. Furthermore, Miwa et al. [14] proposed implementing this method using time domain processing. Furthermore, Miwa et al. [15] uses a two-channel broadband window (center frequency r
+, rz>, find the θth, 1st, and 2nd moments as spectral parameters for each channel, and further find the Parian Σ2° for each channel under the center frequency, and from these, α. We proposed a method to obtain m, n, (k). [Problems with the Prior Art and Objects of the Invention] In summary, the spectral parameters include the θ, 1st, 2nd, and 3rd moments of the power spectrum and/or the normalized moments derived therefrom, f, Σ2.
It can be seen that the number of channels is increased as necessary by focusing on .... For Mo, up to β, n, for Me, M, up to β, MO, MI. For Mz (M3), extraction methods for β, n, and Q', m, and n have already been reported. However, none of these methods provides a method for dealing with the case where the approximate formula (or theoretical formula) becomes more accurate in the future and the types of media parameters increase. Although the Me, Ml, Mz (M3) method can deal with a large number of parameters, the higher-order moments such as M, (M3) change sharply due to minute fluctuations and noise near the edges of the spectral band. do. This has the drawback that the obtained measured values are unstable and errors are easily introduced. On the other hand, the M0 method allows stable measurement values to be obtained, but has the drawback that it requires a large number of channels and the measurement system is complex, large, and expensive. In the present invention, in addition to β, β
, n, or z, m, n, and further provides a method for determining M0 method, M, , M, method 2M0゜Ml, Mz (Ms
) method to provide a method for dealing with future increases in the number of parameter types. [Object of the present invention] An object of the present invention is to obtain spectral parameters that describe the power spectrum of a reflected signal in two or more channels, and to calculate each characteristic value that describes the medium characteristics (experimental formula or theoretical formula). When determining the value of the type, stable measurement possible,
By using the power spectral moments Me and Ml, the number of channels can be reduced as much as possible while ensuring accuracy, and a medium characteristic measurement method capable of responding to future increases and changes in the types of media characteristics is provided. It is in. The present invention focuses on the 0-order and 1-order parameters that can be stably measured among the spectral parameters that describe the power spectrum of the reflected signal.
Focusing on the next moment, an arbitrary number T included in the medium characteristics
The characteristic value (which varies depending on the development of experimental and theoretical formulas) is measured at r/2 or more channels (measurement system in a specific frequency band). By extracting from two types of moments of first order. It is possible to determine any type of characteristic value with the minimum number of channels. In addition, even when focusing on only the 0th moment, 3 types of θth, 1st, 2nd, or higher order moments, it is possible to handle an increase in the number of characteristic values by providing an appropriate number of channels. That is. [Explanation of the principle of the invention] The case of a living body as a medium will be explained. There are currently various experimental and theoretical formulas that describe media characteristics, and considering future developments, it is not appropriate to specify them.
When α in equation (5) is (2) or (2-2), r is (
4) The case of formula will be explained. In other cases, especially when the number of types of medium characteristic values increases, the expansion method will be explained so that it can be easily inferred. First, in the general expression of equation (5), suppose that defeat α2). Substituted in place of f. When α is Taylor-expanded around y=, 0 and up to the yZ term, f Qtα −(4(Jon, vudz) −f) (8), 3
y 2 However, two cases where α is specific are shown here. i) When αxzf −Vcrf+(1+y) +(
m is a function of 2), ii) m child 1. When i=β, r/b = Expln in(r/b) and similarly In(r. Let's consider a specific case. If r/b = f = Tr (1+ y), then r/b=Exp (1nf7 +ny-nF)
(9-1) (6) formula and (8-2) (8-3),
Substituting the approximate equation (9-1) into equation (5) results in a Gaussian distribution equation again. That is, 1 Pa<tm)l” ”
Ak+g)'EXp(#V-Be-Bay-B, f
+1n f 7 + ny −nf )T′ = A −k (1) f! xp (-(-→+B, +n
) fl 2 (B+ -n) )' (Be n ln ding+)
)−(B+ −n) y+−”−←(CBon I
n below the moon) f. Further transforming the inside of () on the right side, we get Σ′ ()=(y+Go?(B+n))” Σ′ − [−2(B+ n) ) ” (F Σ− ”2 (t (Bon1nf1) from,
Furthermore, (B+, n)” Bo +n lnf, )
・The second Exp term is Σ6 However, f r<m>= f r (i(B+
n)”' l Pa (fg) l' D (I
l+ fl)・(Bl n)” Bo +n 1
nT+ ) (10-1) (10-2) Formula r□
is the new center frequency, and Equation (10-3) Σ is the new dispersion, both of which are actually measured spectral parameters. That is, it is calculated from the power spectrum using the following formula. = j! L(12) O where the equation (11) is, the denominator of equation (12) is the 0th moment of the power spectrum, and the numerator of the equation is the 1st moment)
+ The numerator of equation (12) is the second-order moment m2 around the center frequency, and the so-called second-order moment M2 around the origin is (13
), so Mz −J f ” l P−<t, m> l ” d
f (13) The following relationship exists. Therefore, Σ1 may be obtained from the following equation. Σ? =f (M=)2 Machi M・ ("2') However, when calculating Σ using equation (12) or equation (13), the integral range is theoretically -■ to ■, or O~
It needs to be Q. However, this is not practicable, and the range of integration must be kept within a finite range. For Gaussian distribution, it is preferable to limit it to around ±3Σ, but since the exact center frequency is unknown, the integration range shifts left and right. In addition, the spectral intensity is extremely small in the vicinity, and is on the same level as the noise component. When multiplied by (f-T)t, this noise at the base is emphasized and integrated. Σ1 causes a large error due to noise and uncertainty in the integration range and exhibits an unstable value. Therefore, let us consider obtaining the medium characteristics only from M, ,M, without using Σ. Eliminate Σ1 from equations (10-2) and (10-3). Complete i (g) = Complete (1 yen +Bz+n)-'σ1 ・(B+ n)
(14) (complete + fzg+) (”2 +Bz+n)
σ= B, -n
(14') Also, the QOI2 to 0 equation moment M0 is as follows: The number of unknowns in equation (16) is α, In, n, r = 4, and the number of spectral parameters of M o, M + is q = 2. Therefore, the number of channels is p = −
This can be solved by L-=2, ie, by providing a second channel. For the second channel transmission power spectrum (f2.γ2), the reception power spectrum CTt (-1, Z z),
When the moment ``Mo + ''Ml is actually measured, a, m, and n can be determined as unknowns by the actually measured Mo + Ml l ``M(l l ''Ml. Also, k was deleted from MO and Ml (14) expression, as well”
A similar formula (17) in which k is deleted from M,, zM, and (8) formula (%) in which k is deleted from Mo (16) formula of the first channel and 2M0 of the second channel is obtained. (14), (17), and (18) all have unknowns a', m, and n when α=cx fl'', so each unknown can be determined from these three equations. It is possible.B,,B,,B,includes an integral by 2, and it is difficult to find each unknown quantity analytically, but it can be solved by assuming that 72m is constant during a minute interval Δ2 from z=Q, Furthermore, by solving in the same manner for the next section 2=Δ2 to 2Δ2, a', m, and n can be numerically determined sequentially to any depth.Such calculations can be performed using a computer. In a simpler case, when α=βf (m=l, ot=β), the unknown quantity decreases by one, so the two equations (14), (
Equation 17) may be used, and Bll + of Equation (8-3) may be used.
Bl rB! By substituting , we obtain equations (19) and (20). (f r f +(s+)(= +4 r l sβ
dz+n )αI −47, J”βdz −n (19)(
f l-f zg+) (4+4 f 2 , /'βdz
+n) −, iT, J″′βdz −n
(20) T1. α,,Ttl α2 is known,T,. )
, Tt (111 is measured as a function of 2. From this, fβdz and n are found as a function of 2, and β is f
It can be obtained by differentiating βdz by 2. From the above, it can be seen that when the number of unknown types increases in the future, it will be possible to respond by increasing the number of channels. [Embodiment of the invention] For biological diagnosis, T, 3.5 MHz,
Approximately 2 MHz is appropriate for T2 in view of accuracy, resolution, required measurement temperature, etc. The comparison band for α寞 and α2 is 5.
If 0%, 0.74 MHz, 0.42 MHz
values of about 100 psi are possible with current technology using piezoelectric materials such as PZT. To accurately scan the same scanning line, the first channel transducer and the second channel transducer are alternately replaced and measured twice. With PZT using multilayer matching layers, it is possible to create a transducer with a specific bandwidth of about 70%, and with organic piezoelectric materials such as PVF, a transducer with a wider band can be realized. In this case, for one wideband pulse transmission/reception, 3.5MHz and 3.5MHz, respectively.
Two channels of received signals can be obtained using a bandpass filter centered at 2 MHz. The respective bands may overlap slightly. A block diagram of the device in this case is shown in the figure. 1 is a living body to be measured; 2 is a PvF transducer closely attached to the body surface of the living body via jelly, etc.;
4 is a broadband pulse drive circuit; 3 is a receiving amplifier; 2.3.
A transmission/reception switching circuit is provided at the connection point 4, but it is omitted from the figure. 5 is a bandpass filter, 5-1 is a center 3
．． 5 MHz, band approximately 1.8 MHz, 5-2 has band pass characteristics with center 2 MHz, band approximately I MHz,
The respective outputs become the first channel and the second channel. 6-1゜6-2 are respectively the center frequencies "rt + 1"
TZ (A/D conversion 2-hour gate selection with Rino extraction circuit,
FFT, Mo. M1 calculation circuit, f, l'' division circuit on the center side M. It can be configured digitally using a microcomputer, etc. to calculate the wave, or it can be configured by using a quadrature detector to calculate the in-phase output I and quadrature output Q. Furthermore, I and Q are obtained by time differentiation, and then the center frequency f, = IQ - IQ I2 + Q2 is obtained using a circuit that calculates the following formula from I, Q, I, and Q. All real-time calculations are performed using analog circuits. 7 is the output Tl of 6-1 (from Kl and the output T'l<m> of 6-2, the two-dimensional simultaneous algebraic equations of equations (19) and (20) are n, fβdz It is a calculation circuit that outputs , and can be configured either digitally or analogously. 8 is a time differentiator circuit that differentiates βdz with time. Time is related to depth at t-22/C (C - speed of sound) The time differentiation is synonymous with the differentiation by 2. The output of 8 gives the z (t) direction distribution of β. 7 is a display control circuit that displays the distribution diagram of n and/or β in two directions. When the transducer is moved in the X direction by a scanning mechanism (not shown), n and/or β on the cross section in the X2 plane
distribution image can be displayed. When the circuit is configured entirely in analog fashion, it becomes time domain processing and can be displayed in real time. Next, a modified example will be shown. For the three unknowns k, β, and n, M
An energy ratio method that focuses only on o (q = 1) has been reported using 3 channels, but for 4 unknowns (r = 4) or more such as k, a, m, and n, only Mo It is easily understood that the problem can be solved by focusing on P≧4 and using four or more channels. However, it has the disadvantage of increasing the number of channels. In addition, a method focusing on Me, M+, and Mt has been reported to use two channels to obtain k, τ, m, n, (γ=4), but in the case of T≧5, two channels or It can be solved using the above. Also, focusing on a total of q (2≧4) higher-order moments of Me, MI, Mz, M's or higher, and using a channel of P≧γ/q (P=integer) for the unknown γ, γ unknowns can be determined. However, it cannot escape the drawback of instability of higher-order moments. Practically speaking, MO, M, law, or MO, M', , PJ
The It method is convenient. [Effects of the Invention] According to the present invention, by appropriately selecting the spectral parameters and the number of channels, it is possible to cope with future fluctuations in the number of parameters in the medium characteristic near-value equation or the theoretical equation. This has the effect of optimizing the system in terms of both parameter stability and minimizing the number of channels. Reference list (1) S, A, Goss, R, L, John
ston and F, Dumr” Compreh
comprehensive compilation of emp
iricalultrasonic property
es of maa + analian tissues''
, Journal of the Acousti
calSociety of America, vo
l 64+ 1978+pp 423-457 (2) P, A, Naraya and J, 0
phir″On the frequency dep
end of attenuation in
normal and fat 1iver'
IEEE+5onics and ultrason
ics+ pp 379-383+ SO3ONo 6
．． Nov, 1983 (3) J, C, Bamber and C,R,)
Fill” Acoustic properties
of normal and cancerous h
uman 1iver 'IJlrasound in
Med & Riot, Vol 7. PP 121
～133, 1981゜(4) D, N1cholas ”Evaluation of backscat
ering coefficients for ex
cised human tissues:
results, 1nter-pretatio
n and associated measurements
ents ”Ultrasound in Me
d & Biol, Vol 81 Nl 1
pp17-28 1982゜[5] Mitsuhiro Ueda, Yasuhiko Ozawa Spectrum of echoes scattered in irregular media US 82
-42. Institute of Electronics and Communication Engineers, 1982.10.18M, U
eda and Y, Ozawa″5pectra
l analysis of echoes forb
ackscattering coefficient
lIleasurement, ”J, Acous
tic 5ociety of America. [6] Miwa, Shimura, Murakami Specimen identification method % formula % Method for measuring biological tissue characteristics using ultrasound Patent application 1987-57573゜ (9) Roman Kuc, M, Schwart
z and Lajos Von1csky ” Parameter fc estimation o
f the acousticattenuation
coefficient 5lope for 5o
fttissue” 1976+ Ultrasonics Symp, P
roceeding, IEEE Cot, I 76,
CHH20-5Sυ [10] Miwa et al. Ultrasonic measurement method and its device Patent application 1987-65536 [11] Nakayama and Shiba ultrasonic measurement method and its device Patent application 1987-77226 [12] Miwa et al. measurement using spectral shape Apparatus Patent Application No. 58-45396 [13] Miwa, Shiba Power, Measuring Device Using Spectrum Moment Patent Application No. 58-142893 [12] Miwa et al. Patent Application No. 1987-85210

[Brief explanation of the drawing]

The figure is a block diagram of an embodiment of the present invention.

Claims (1)

[Claims] In measuring ultrasonic characteristic values of a medium by transmitting ultrasonic pulses into the medium and receiving and analyzing the reflected waves, a measurement channel operating in a plurality of spectral bands is provided. For each channel, determine the spectral parameters that describe the transmit spectrum and receive spectrum,
An ultrasonic medium characteristic value measuring device for further extracting medium characteristic values from these parameters, characterized by having at least the following three means. (1) The number of channels is p, the number of types of spectrum parameters is q, and the number of types of medium characteristic values included in the transfer function of the medium that converts transmitted waves into received reflected waves is r (p, q, r are means to provide the number of channels that satisfies p≧r/q when r is 3 or more (2) 0th and 1st order of power spectrum as types of spectral parameters (3) means for extracting the moment or its normalized moment; (3) means for calculating medium characteristic values from the extracted spectral parameters;