Method and device for generating images from ultrasonic signals obtained by echography

• US 4,729,019 A
• Filed: 10/15/1986
• Issued: 03/01/1988
• Est. Priority Date: 10/18/1985
• Status: Expired due to Fees

• Current Assignee: Cgr Ultrasonic
• Original Assignee: Cgr Ultrasonic
• Inventors: Rouvrais, Bernard
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US06/918,883
• Time in Patent Office: 503 Days
• Field of Search: 358/112, 358/96, 358/166, 358/167, 382/6, 364/414, 128/653, 128/660, 128/661
• US Class Current: 348/163
• CPC Class Codes: G01S 7/5206 Two-dimensional coordinated display of distance and direction; B-scan display
• CPC Class Codes: G01S 7/52073 Production of cursor lines, markers or indicia by electronic means
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: H03H 17/0202 Two or more dimensional filters; Filters for complex signals multidimensional convolutions G06F17/153

Abstract

The method and the device provides for the automatic processing of digitalized signals in order to

remove speckle from the said signals by comparing the digitalized signal with a noise model obtained after learning its statistical properties,

extract, from the signals, echos not characteristic of the contour of the organ to be displayed,

extract the characteristic parameters of the contour from the said signals,

calculate the contour from these parameters by likening them to a pre-established model of variable parameters,

combine the digitalized signal, before or after eliminating noise, with the signal that is characteristic of the estimated position of the contour,

display the resultant signal in which the contour appears as an overlay.

Claims

1. Method for generating images from source ultrasonic signals obtained by echography, wherein the echos signal is sampled and digitalized to form a succession of samples x_i,j^t as a echographical signal with a coded digital value of luminance X at an instant t, identified in the image by the row i of the horizontal column and the row j of the vertical column to which they belong, wherein said echographical signal is automatically processed to superimpose, on the image, a contour of the analyzed organ by the following steps:

reducing the speckle present in said echographical signal through filtration by generating a prediction divergence signal which yields useful data by taking the difference between the digitalized signal x_i,j^t and a prediction signal calculated from a model of speckle, this step of reducing further providing a filtered digital signal F_i,j^t and a binary signal B_i,j^t deduced from the filtered signal produced by said filtration, at state 0 for the points of the filtered signal that have zero luminance and at state 1 for the other points;

processing said binary signal to extract from it the data corresponding to echos recognized as being non-characteristic of the contour, and outputting a signal I_i,j^t ;

processing said signal I_i,j^t by bridging to minimize discontinuities by eliminating isolated pixels, and outputting a bridged signal I_i,j^t ;

processing said bridged signal I_i,j^t to extract from it the characteristics parameters that can be used in a model of the contour of the organ examined, on the basis of a zone of maximum area, called a zero density zone, which comprises only zeros and corresponds to the interior of the organ analyzed;

calculating the pixels of the contour on the basis of said extracted characteristic parameters and the equation of the model of the organ;

replacing the digital signal, before filtration or when it is filtered, by an overlay signal for all the pixels of the contour;

displaying the result of said step of replacing the digital signal by said overlay signal.

2. Method according to the claim 1 wherein, for the processing of signals obtained by cardiac echography, in which echos are significant of the position of the mitre valve and non-characteristic of the cardiac contour, the processing of the binary signal in the second stage consists in performing the logic AND operation between two binary signals B_i,j^t and the same signal q images before B_i,j^t-qT.sbsp.i these two signals corresponding to the farthest positions of the mitre valve in semi-periodical movement substantially equal to qT_i, the entire value of a q being readjusted, if necessary, according to the cardiac rhythm.

3. Method according to claim 1 wherein the signal I_i,j^t is bridged by eliminating the isolated pixels at logic level 0, isolated in the sense that they are non-contiguous along a distance equal at least L pixels.

4. Method according to claim 1 wherein the speckle is considered to be a circular noise, the noise model used for generating the prediction signal from the neighbouring samples being a one-dimensional self-regressive model, the order and the coefficients of which are pre-determined by a preliminary learning of the statistical properties of noise in a sequence of images used as a reference.

5. Method according to the claim 4 wherein the preliminary learning of the properties of the noise is done by measuring its mean value, its variance and its self-correlation function on the basis of which the order and coefficients of the noise model will be fixed.

6. Method according to the claim 4 wherein for the processing of signals obtained by cardiac echography, the statistical properties of a noise in a sequence of reference images are estimated in a zone corresponding to the interior of a ventricle where the graininess of the image is likened to the speckle.

7. Method according to the claim 1 wherein, in the fourth stage, the search for the zero density zone with the maximum area is obtained by successive reductions of a zone with a simple geometrical shape, definitely including the image of the analyzed organ, until the density measured in the reduced zone is zero, the geometrical shape selected being capable of being registered in the contour model of the analyzed organ, this zero density zone being characterized by a position vector P at each image.

8. Method according to the claim 7 wherein, for the processing of signals obtained by cardiac echography, the model of the contour is an ellipse and the geometrical shape chosen is a square, the successive reductions being obtained by a reduction of the side of the square at each stage in the search.

9. Method according to the claim 8 wherein, at each stage of the search, the preceding square is subdivided into four squares of the same dimension, reduced in relation to the dimension of the preceding square, these squares partly overlapping one another, the square with the minimum density being selected for the following stage.

10. Method according to any one of the claims 7 to 9 wherein the position vector P is corrected if necessary on the basis of the position vector associated with the preceding image P^t-T.sbsp.i where T_i is the image period, to ensure a sequential follow-up of the contour, the resultant vector being a vector P.

11. Method according to the claim 10 wherein the parameters in the contour model, deduced from the coordinates of the vector corrected according to the coordinates of the vertices of the zero density square zone, P(xHD 1, x_n, Y₁, Y_n), are on the one hand, the center of this zone with coordinates ##EQU12## on the other hand, two radii ##EQU13## and ##EQU14## and λ

a parameter for adjusting the ellipse such that ##EQU15## λ

_min being a predetermined value of less than 1, the contour being likened to an ellipse defined by all the pixels with coordinates x and y such that;

space="preserve" listing-type="equation">R.sub.min.sup.2 <

λ

.sup.2 (x-g.sub.x).sup.2 +(y-g.sub.y).sup.2 <

R.sub.max.sup.2.

12. Device for generating images, using ultrasonic signals obtained by echography, comprising:

a memory input circuit a sequence of digitalized samples X_i,j^t ;

a speckle filtering circuit connected to said memory input circuit and providing an initial output to give a filtered digital signal of the speckle F_i,j^t, and a second output to give an associated binary signal that has the logic 0 level for those points with a filtered value of zero and the logic 1 level for all the other points,

a processing channel comprising, in series, a circuit for extracting data corresponding to echos not characteristic of the contour linked to the second output of the filtering circuit, bridging circuit to eliminate pixels detected as being isolated, a circuit for extracting the parameters of the contour and a circuit to calculate the contour,

a single direct channel comprising a change-over switch, the inputs of which are linked to the output of the input circuit and to the first output of the filtering circuit, and the output of which transmits, to a delaying circuit, either the succession of digitalized samples X_i,j^t or the filtered sequence F_i,j^t, wherein the output of the direct channel after the delaying circuit and the output of the processing channel after the contour-calculating circuit is linked to the inputs of a circuit for overlaying the contour on the image, the output of this overlaying circuit being connected to a visual display screen.

13. Device according to the claim 12 wherein the filtering circuit comprises:

a predictor circuit to calculate a prediction value X_i,j^t on the basis of the preceding samples X_i,j^t and a noise model followed by the predictor,

a calculation circuit to calculate a prediction divergence e_i,j^t between the prediction value X_i,j^t and the real value X_i,j^t,

a multiplier circuit to multiply by k₂ to perform the operation k₂ e_i,j^t,

two comparison circuits connected to said multiplier circuit, the first comparison circuit limiting the signal k₂ e_i,j^t to the maximum range and giving the filtered signal F_i,j^t, and the second comparison circuit giving the binary signal B_i,j^t at level 0 for k₂ e_i,j^t ≦

0 and at level 1 for k₂ e_i,j^t >

0.

14. Device according to claim 12 wherein the contour parameters extracting circuit comprises the following elements connected in series:

a set of memories,

a density calculating circuit indexed by a zone position memory circuit through a change-over switch transmitting the pre-determined position of an initial zone,

and a circuit to calculate the parameters of the contour.