JPH0246215B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an ultrasonic diagnostic apparatus that separates a pulse echo signal from a living body into a specular component that represents the outline of a tissue and a random component that is effective for extracting characteristic parameters inside the tissue. The present invention relates to an apparatus capable of simultaneously displaying shape information and quantitative information based on characteristic parameters in an ultrasonic tomographic image.

Conventional tissue quantification using ultrasound is performed by using the spectral frequency dependence of an echo signal (random) component in a weakly inhomogeneous medium, or by measuring the frequency distribution of the amplitude or time interval of this signal. However, if a large reflected echo signal (spectral) component due to tissue discontinuity is superimposed on this signal, an error will occur in the above quantification.

Furthermore, in conventional B-mode images, the specular component and the random component are not separated and are displayed in an overlapping manner, so there is a drawback that tissue boundaries cannot be clearly defined.

An object of the present invention is to separate and process a pal-echo signal from a living body into a specular component that represents the contour of the tissue and a random component that is effective for extracting characteristic parameters inside the tissue. An object of the present invention is to provide a device capable of simultaneously displaying quantitative information based on characteristic parameters as an image.

The present invention makes it possible to more easily separate specular components and random components by providing a separation circuit and using a combination of not only a linear filter but also a nonlinear filter for the pulse echo signal from a living body as a separation circuit. , the shape and quantitative information of biological tissues can be displayed simultaneously.

Figure 1 shows the entire configuration of the device. The ultrasonic echo signal x(t) from the living body is processed by a separation circuit consisting of a linear filter and a nonlinear filter, for example.
It is separated into a large echo signal (specular) component S(t) that returns from a clear tissue discontinuity in the living body and a weak echo signal (random) component R(t) that returns from a fine weak heterogeneous structure of the tissue. . By using a combination of a linear filter and a nonlinear filter, this separation can be performed more easily and accurately.

The tissue contour is clearly defined from the separated specular component S(t), and the specular component S(t) and/or
Alternatively, feature parameters inside the tissue are extracted from the random component R(t). This operation is performed two-dimensionally to display an image of the tissue outline and characteristic parameters at the same time.

FIG. 2 shows a block diagram of the separation circuit in FIG. This is an example in which individual filters are combined using a linear filter, a limited inverse convolution filter and a convolution filter, and a non-linear filter using a non-linear matching filter. echo signal x(t)
is the specular component S(t) and the random component R(t)
It is expressed as the sum of That is, it can be expressed as x(t)=S(t)+R(t) (1), where S(t) is ideally the impulse response h(t) of the resonator determined by the electrical characteristics, and the delay time τ _i and amplitude ai as parameters, S(t)=〓 ⁱ aih(t-τ _i )=[〓 ⁱ aiδ(t-τ _i ) _(*) h(t) (2). However, ○* indicates convolution, and δ(t) indicates a delta function. Conventionally, the specular component S is extracted from the received signal x(t).
In order to find (t), for example, the correlation between the impulse response (t) of the vibrator and x(t) is found, and it is assumed that a specular reflector exists at a position where a large peak value is exhibited. When determining the position of this specular reflector, rather than using normal linear correlation, it is better to use a nonlinear matching filter whose output is the inverse of the error power between the reference function and the input function (when a match is achieved, The error becomes almost zero, giving a very large output, and if there is even a slight deviation from the matching point, the error increases and the output becomes smaller), resulting in a very sharp peak in the output. This will improve accuracy. However, when using a nonlinear filter, the principle of superposition of linear systems does not apply, and if two or more specular reflected waves that overlap, even partially, are input, it may not be possible to separate them well. This causes problems such as a significant drop in the output of the filter for each specular reflected wave. Therefore, as shown in claim 1, first, the received signal x(t) is deconvolved (linear filter operation) using the impulse response of the vibrator, and x(t) is
The nonlinear filtering described above may be performed after reducing the overlap between the specular reflected waves as much as possible by shortening the pulse duration of .

An example of a procedure for estimating a _i and τ _i in equation (2) will be shown below.

(i) Taking into account the band limit of the vibrator, obtain the estimated value g(t) of the impulse response using restricted inverse convolution assuming that the noise variance is known.

(ii) The signal obtained by discarding the low-level waveform of this g(t) signal is passed through a nonlinear matching filter as shown below to obtain the sign sgn(a _i ) of a _i and τ _i .

The nonlinear matching filter described here calculates the delta function when the band is limited according to the band of the resonator, and then uses the two delta functions δ ⁺ (t) and δ ^- (t) with different polarities as the standard. The signal used is as follows.

The lower level is the impulse response g(t) which is the output of the truncated inverse convolution filter and the reciprocal of the following two error powers within an appropriate interval Δ N ⁺ ,
Calculate N ^- .

N ⁺ (τ _i ) =1/∫〓｜kg(t)−δ ⁺ (t−τ _i )｜ ² dt (3) N ⁻ (τ _i ) =1/∫〓｜kg(t)−δ ⁻ (t−τ _i ) | ² dt (4) where k is the impulse response g(t) within the interval Δ
is a normalization constant that equalizes the maximum amplitude of the delta function. While shifting τ _i within the interval, N ⁺ (τ _i ),
When calculating N ^- (τ _i ), when the correlation between the impulse response and the delta function is high, one or the other will give a sharp peak. A threshold value is detected, and the peak position at this time becomes τ _i . However, the symbol "^" indicates an estimated value. The sign of sgn(a^ _i ) is determined by whether N ⁺ or N ^- gives a peak, respectively, plus or minus.

(iii) The square of the error between the composite value by convolution { 〓 ⁱ (t−τ^ _i ) | a^ _i | agn (a^ _i )}○*h(t) and the input signal x(t) is the minimum. Find adaptive |a^ _i | such that This operation is performed within a certain gated time window.

(iv) The composite value of a^ _i that minimizes the square of the error gives the specular component S^(t), and the error x
(t)−S^(t) is the random component R^(t) in equation (1)
For that, R^(t)=x(t)−S^(t) (5).

FIG. 3 shows an example of the filter output of an echo signal obtained for 3 mm thick acrylic acid. The vibrator used has a center frequency of 2.25MHz. Figure A shows the original waveform reflected from the front and back surfaces of the acrylic board. B shows the output (impulse response of the acrylic plate) after the original waveform is passed through a deconvolution filter. C is the result of passing this impulse response further through a nonlinear matching filter. Two sharp peaks corresponding to the front and back surfaces of the acrylic plate were obtained.

FIG. 4 shows the processing results of echo signals for chicken meat. Similarly, a vibrator with a center frequency of 2.25MHz was used. A is the original waveform. B is its impulse response (output of the deconvolution filter). C indicates a delta function when the band is limited according to the band of the vibrator used, and this is used as a reference signal for the nonlinear matching filter. D is the output result of the nonlinear matched filter. E and F indicate the finally separated specular component and random component, respectively.

In Fig. 1, the specular component is used to clearly show the contour of the tissue, but since the impulse response from the tissue or the output of the nonlinear matching filter is used to obtain the specular component, this output is When applied to contour display, it is possible to create images with higher distance resolution than envelope detection.

According to the present invention, a pulse echo signal from a living body can be separated into a specular component representing the contour of the tissue and a random component effective for tissue characterization such as extraction of characteristic parameters inside the tissue. It enables simultaneous image display of shape information and quantitative information based on characteristic parameters.

[Brief explanation of drawings]

FIG. 1 shows the entire configuration of the device. FIG. 2 is a block diagram of the signal separation system. FIG. 3 shows an example of a filter output of an echo signal obtained for an acrylic plate. FIG. 4 shows the processing results of echo signals for chicken meat.

Claims (1)

[Claims] 1. Ultrasonic waves are transmitted to a living body, and the ultrasound echo signals received from the living body are divided into large echo signal (spectral) components returned from distinct tissue discontinuities within the living body, and microscopic components of the tissue. a circuit that separates a weak echo signal (random) component returned from a weakly non-uniform structure, the circuit comprising a combination of a linear filter and a non-linear matching filter that outputs the reciprocal of the error power in the matching section; An ultrasonic diagnostic apparatus characterized by having a display circuit that extracts and displays tissue characteristic parameters based on the component and/or the random component. 2. Claim 1, characterized in that the amplitude of the specular component obtained from the output of the nonlinear matching filter is optimized so that the error is minimized with respect to the received ultrasonic echo signal. The ultrasonic signaling device described. 3. The ultrasonic diagnostic apparatus according to claim 1, wherein the display circuit includes a circuit for two-dimensionally displaying a tissue contour based on the specular component.