WO2010018282A1

WO2010018282A1 - Method and apparatus for correcting ultrasound images by means of phase analysis

Info

Publication number: WO2010018282A1
Application number: PCT/ES2009/070303
Authority: WO
Inventors: Carlos Fritsch Yusta; Montserrat Parrilla Romero; Jorge CAMACHO SOSA DÍAS
Original assignee: Consejo Superior De Investigaciones Científicas (Csic)
Priority date: 2008-08-08
Filing date: 2009-07-23
Publication date: 2010-02-18
Also published as: ES2332637A1; ES2332637B1

Abstract

The invention relates to a method and apparatus enabling the real-time correction and improvement of the quality of ultrasound images obtained using traditional methods. For this purpose, the invention consists of multiplying the formed samples by phase coherence factors (CF(κ)) based on an analysis of the dispersion of the phases of the aperture data (S _i )(κ)).

Description

PROCEDURE AND APPARATUS FOR THE CORRECTION OF ULTRASONIC IMAGES BY PHASE ANALYSIS

D E S C R I P C I Ó N

OBJECT OF THE INVENTION

The object of this invention is to provide a method for correcting and improving, in real time, the quality of the ultrasonic images obtained by conventional methods by means of arrays or arrays of transducer elements and digital beam shaping techniques. In addition, an apparatus for carrying out the process of the invention is also described, and which can be easily incorporated and adapted to existing beam shapers.

BACKGROUND OF THE INVENTION

Conventional ultrasonic imaging systems are based on a set or array of N transducer elements, usually piezoelectric, that emit ultrasonic pulses in the direction of the medium to be inspected, where the moment of emission of the pulses is timed so that the pulses Individuals are added forming an ultrasonic beam. A focal law is the set of delays calculated to deflect and focus the beam in a certain direction and range depending on the geometry of the array, the coupling to the medium to be inspected and the propagation speeds of the ultrasound. Changing the focal law modifies the direction and focal length of the beam, which allows sweeping the region of interest with foci located at the same or at different depths. With linear arrays the scanning is flat and, with the two-dimensional scanning of a volume can be performed. On reception, the echoes that reach the receiving array (usually the same as the emitter) as a result of reflections in discontinuities in the medium, are amplified, digitized and delayed individually for each element / N of the receiving array, 1≤i≤ N, obtaining the data of the aperture S, (7cJ, where k represents an index in the signal of length L (1≤k≤L). The focal law applied in reception compensates for differences in the flight time of the ultrasound from Ia emission to the focus and to each element When adding the N opening data (previously delayed), constructive interferences occur if they come from the focus or destructive if they come from other regions, a process called coherent sum.The most advanced systems dynamically vary the focal law so that the focus is placed, at all times, on the position occupied by the ultrasonic pulse in its propagation through the medium, thus obtaining an image focused on its entire length ( dynamic focusing technique).

When visualizing the intensity of the received signals, the image shows the amplitude of the reflectors in the positions they occupy. The element that performs the focusing of the received signals is called a beam shaper. Methods for performing conformators with dynamic focusing are described, for example, in C. Fritsch et al., "Consistent composition of signals by progressive focal correction", Pat. 2004/00203, 30 Jan. 2004, or in M. D. Poland, "Ultrasonic diagnostic imaging with automatic adjustment of beamforming parameters", US2007 / 0088213 A1, Apr. 19, 2007.

It is well known (for example, GS Kino, "Acoustic waves: devices, imaging and analog signal processing", Prentice Hall Inc., 1987) that the quality of the ultrasonic images obtained with a beam former as described above is mainly limited , by: a) The lateral resolution or ability to distinguish two nearby reflectors from each other.

b) The dynamic range or ratio between the highest and lowest intensity signals detectable over the background noise without saturation, which in turn limits the contrast.

c) The presence of artifacts and, in particular, of grid lobes that appear when the effective distance between elements is greater than half a wavelength. The limit of distance between elements less than λ / 2 is frequently exceeded with two-dimensional arrays, to maintain electronic complexity in reasonable dimensions and also in image applications for Non-Destructive Evaluation (END).

The lateral and grid lobes significantly deteriorate the quality of the image obtained. Both produce indications where there are no reflectors, limiting the dynamic range and contrast of the images. In particular, anechoic areas where indications should not appear are contaminated by those corresponding to the lateral or grid lobes of nearby reflectors or dispersers.

On the other hand, the width of the main lobe in the lateral pattern of the beam primarily determines the lateral resolution of the image for intense signals. For the weakest, the lateral resolution is determined by the width of the lateral lobes on both sides of the main. The apodization techniques standardize the lateral resolution, reducing the amplitude of the lateral lobes at the expense of widening the main lobe, with the consequent loss of lateral resolution for intense signals. d) The appearance of phase aberrations, which are produced by the variations in ultrasonic propagation speed in non-homogeneous media, causing focalization errors when modifying the trajectory and / or the flight time of the ultrasonic pulse. These focusing errors blur the image and cause loss of resolution and contrast.

The first three limitations are a function of the geometry of the array and the wavelength. Thus, for a long time, it was considered that these limits could not be exceeded. However, more recently, techniques have been proposed that allow the introduction of corrections in the images obtained, with the aim of reducing or canceling unwanted indications in the image, produced by lateral or grid lobes, maintaining or improving other aspects such as resolution. lateral and the signal-to-noise ratio.

The idea generally followed so far is to estimate a coherence factor indicative of the quality of the targeting for each sample k conformed. A shaped sample is the result of adding the signals received once focused. Thus, when applying the coherence factor to the output of the shaper, samples with a high coherence value are maintained, while those with a low coherence value are reduced.

In the original proposal (KW Rigby, "Method and apparatus for coherence filtering of ultrasound images", US Pat. 5,910,115, Jun. 8, 1999), for each rank k, the coherence factor C (TcJ a from the data of the opening S, (k), λ ≤i≤N, as the ratio between the absolute value of the coherent sum and the incoherent sum, that is:

A variant (K. W. Hollman et al., "Coherence factor of speckle from a multi-row I tried", Proc. IEEE Ultrasonic Symposium, pp. 1257-1260, 1999) is:

where energies are related instead of amplitudes. Another variant (A. L. Hall et al., "Method and apparatus for coherent imaging", US Pat. 6.071240, 6 Jun. 2000) relates the sum consistent with that obtained by a second shaper with focalization delays equal to zero. In this case, the coherence increases with the dissimilarity between both magnitudes.

The coherence factor can be used to adjust emission or reception parameters in an iterative process and optimize some quality criteria (KF Ustuner et al., "Coherence Factor adaptive ultrasound imaging methods and systems", Pat. US2005 / 0228279, Oct 13. 2005).

Another variant is the generalized coherence factor (P. C. Li, M. Li,

"Adaptive Imaging using the Generalized Coherence Factor", IEEE Trans. Ultr., Ferroelec. and Freq, Contr., 50, 2, pp. 128-141, 2003) that is obtained from the PQM spectrum) of the opening data:

where the parameter M «N, chooses a low frequency band of the spectrum. For M = O the result is equivalent to that of Equation (2). For M> 0, GCF (k) ≥ C (k), which decreases the correction of non-coherent signals, as are the indications of diffuse reflectors.

Recently, a specific method has been proposed to suppress the indications of the grid lobes (K. F. Ustuner et al., "Adaptive grating lobe suppression in ultrasound imaging", US Pat. 7207942 B2, Apr 24, 2007). Using cross-correlation techniques, the method determines whether an indication comes from the focus or from a grid lobe, filtering the data of the aperture or the result of the coherent sum based on the result.

DESCRIPTION OF THE INVENTION

In this document, the term "opening data" refers to the signals received by the transducers, once amplified, digitized and temporarily delayed (focused), as defined above in this document. The opening data, then, are added together to obtain a single value, which will be called a "conformed sample". The element that performs, at least, the targeting of the received signals and the subsequent sum of the opening data is a "beam shaper".

The image quality improvement procedures of the prior art are based on weighing the formed sample using coherence factors obtained from amplitude relations of the opening data, without explicitly considering the important phase information contained in the signals. That is why they fail to cancel or reduce the grid lobes, where the amplitudes of the coherent and incoherent sums are similar, producing a coherence factor close to the unit that leaves the shaped sample virtually invariable and, therefore, keeps the false ones. indications of the grid lobes.

The present invention, on the other hand, overcomes these limitations of conventional procedures by analyzing the variability of the phases in the opening data, evaluating a phase coherence factor with values between 0 and 1. Therefore, from The phases of the opening data calculate a phase coherence factor that is independent of the amplitude of the signals. Thus, while the ratio of amplitudes of other approaches produces coherence factors whose values are of the order of the quotient between the amplitude of the lateral lobe and the main lobe, the process of the invention produces phase coherence factors of lower values, thus achieving a more effective cancellation of inconsistent data.

This new procedure improves the quality of the images obtained by an ultrasonic imaging system in all the aspects defined above. That is, it is possible to reduce the lateral lobes, the grid lobes, the effects of the phase aberrations and the main lobe is narrowed to improve the lateral resolution, contrast and dynamic range of the image and the signal-to-noise ratio.

In addition, the method proposed in this invention allows adjusting the level of the correction, either automatically or by the operator of the system, being able to reach in the limit to cancel all the indications of the lateral lobes and reduce the width of the main lobe, with the consequent increase in lateral resolution and contrast.

Therefore, according to a first aspect of the invention, a method for the correction of ultrasonic images is described, characterized in that it comprises the operation of multiplying the sample formed by a phase coherence factor based on the dispersion of the phases of the opening data.

The term "phase dispersion", in this document, refers to the degree of diversity of the phases of the signals that make up the opening data. For example, a low phase dispersion corresponds to a set of signals that are all in phase or almost in phase or, in other words, that has a high phase similarity. Therefore, the terms "dispersion" and "similarity" of phases are in inverse relationship. In the same way, we will say that a set of signals have a high "phase coherence" if the dispersion of its phases is low or, equivalently, if its similarity is high. It is necessary to emphasize that the term "dispersion", in this context, intends to refer to any parameter that reflects the distribution of the phases, without necessarily being limited to the parameters that are commonly known as "dispersion" in the field of statistics. As will be described later in the present document, there are different ways of quantifying the dispersion of the phases corresponding to different preferred embodiments of the invention.

Therefore, from the analysis of the dispersion of the phases of the echo signals received by the different elements of the array, once amplified, digitized, and delayed to create a focus in a given direction profundidado and a depth r _k , the phase coherence factor, CF (k), with a value between 0 and 1 for each range k, 1≤ k ≤ L. A high value of the phase coherence factor (close to 1) means that the delays applied have correctly compensated for the differences in flight time to each element and, therefore, come from the focus. A reduced value of the phase coherence factor (close to 0) means that the delays applied have not adequately compensated for differences in flight times and, consequently, the signals do not come from the focus. The phase coherence factor is used to weigh the result of the sample formed according to the following expression:

y (k) = CF (k) -χ (k) = CF (k) ∑ S ₁ (k) (4)

so that, when CF (TcJ »1, the result y (k) of the weighting will be basically equal to the simple sum x (k) of the opening data, whereas, when CF (k)« 0, The output and (k) will tend to zero, significantly limiting the amplitude corresponding to samples formed with a low coherence.

Thus, the procedure proposed in this invention recognizes the origin of the signals and acts to correct the unwanted effects on the image or to improve its quality.

Thus, a reflector located in the central direction of the main lobe produces signals in phase, so that the sum is constructive and gives rise to the maximum amplitude in the lateral pattern of the beam. Under these conditions, the phase coherence factor CF (TcJ is maximum, with a unit value (zero phase dispersion).

However, when the reflector moves laterally within the main lobe, some signals are no longer in phase. By increasing Ia phase disparity, the phase coherence factor CF (k) is reduced. Thus, when correcting the sum of the data of the opening by application of (4), the width of the main lobe is reduced and, consequently, the lateral resolution is improved.

If the reflector is placed on a lateral lobe, the signals are no longer constructively composed. The phases of the opening data show a great disparity that produces a very low phase coherence factor, CF (k) ^to O. By multiplying the sum of the opening data by this low value according to (4), they are reduced or cancel the indications of the lateral lobes.

The grid lobes result from distributed replication of indications of reflectors located on the main lobe. With the broadband signals used in ultrasonic imaging, the replicas are partially coherent compositions: only a small fraction of the N opening data is in phase and the rest shows diverse phases. Overall, the dispersion of the phases is high, producing a low value of the phase coherence factor CF (k) = O. By multiplying the sum of the opening data by this value, according to equation (4), the indications of the grid lobes are limited.

Likewise, the signals received when phase aberrations occur show a high phase dispersion, resulting in a low phase coherence factor CF (k) »0. By multiplying the sum of the opening signals by this low value according to (4), these indications are reduced.

The phases of the opening data are obtained from the analytical signal expressed with its components in phase Sl, (k) and quadrature SQ, (k) as: -i SQXk) φ, (k) = tan (5)

YES Xk)

The beam shapers that operate in the baseband directly have the analytical signal in phase and quadrature, so that the application of Equation (5) is enough to obtain the phase φ, (k) in each channel / for each range k . The resulting phase must be in the interval (-π, π).

In case of using a beamformer operating in radiofrequency, however, a previous operation is required to obtain the quadrature signal by means of a Hilbert transformer, whose realization is known (A.V. Oppenheim, R. W. Schafer: "Digital Signal

Processing ", Prentice-Hall, 1975):

SQXk) = Hilbert [SXk)] siχk) = sχk)

There are also other known techniques that allow an approximation to the analytical signal to be obtained through quadrature sampling [K.

Ranganathan et al., "Direct sampled I / Q beamforming for compact and very low-cost ultrasound imaging", IEEE Trans. on Ultrason., Ferroelect.

Freq. Contr., 51, 9, pp. 1082-1094, 2004].

With the described procedures, the instantaneous phase of the opening data can be obtained with arbitrary precision. Greater precision requires greater electronic complexity, but does not provide a corresponding improvement in the suppression of non-coherent signals. For this reason the phases will preferably be calculated with relatively low accuracy, typically with a resolution between 1 and 8 bits. Once the phases of the signals from Equation (5) have been evaluated, the preferred phase coherence coefficient is calculated from the following expression, where the max () function serves to prevent CF (k) from taking negative values :

CF (Jc) = max (0, l - a- f [φ, (k)]) (7)

where:

Ψ (k) ^is' f ^ase of the sample k of the signal / data of the opening; a is an adjustment parameter; Y

^J is an estimator of the dispersion (or similarity) of the phases of the signals of the opening. It can be a dispersion measurement statistic, such as the range, the

2 standard deviation ( ^σ ), the variance ( ^σ ), the kurtosis, etc. Alternatively, similarity measurement functions may be chosen, in which f () decreases with increments in the similarity of the phases.

From the definition given by Equation (7), the values of CF (k) will be between 0 and 1. The maximum CF (k) = 1 is obtained when

JW ⁱ \) \ = o, that is, when the dispersion of the phases is minimal or, equivalently, their similarity is maximum. The minimum CF (k) = 0 is obtained when ^{• /} l ^ v ^ ll> 1 / _Qi that is, when the dispersion of the phases reaches a certain programmable value by means of the α parameter.

In a preferred embodiment of the invention, the standard deviation J is used as estimator. In this case, the coherence factor It is calculated as:

CF _x (Jc) = max (0, 1 - a, .σ [φ, (Jc)]) (8)

where cr [<p, (£)] represents the standard deviation of the phases of the opening data, and the normalization coefficient CU ₁ preferably takes

values between O and -. π

Equation (8) has a unit value when all phases are equal, and therefore the standard deviation is zero, and a zero value when

The standard deviation of the phases reaches or exceeds the value -.

In another preferred embodiment of the invention, the variance is used as estimator ^J :

CF ₂ (Jc) = max (0, 1 - a ₂ σ ² [φ, (Jc)]) (9)

where σ ² [#>, (&)] represents the variance of the phases of the opening data, and the normalization coefficient at ₂ takes, preferably,

values between 0 and π

Similarly, equations can be defined to calculate the phase coherence coefficient based on other statistical moments that measure the dispersion of the random variable φ _t (k), such as kurtosis or other higher order moments.

In a particularly interesting preferred embodiment, the phases of the signals of the opening data, which occupy the interval (-π, π), are evaluated with a single bit b to which the values -1 and +1 are assigned for the angular intervals (0, π] and [ -π, 0], respectively.In this situation the electronic complexity is minimal, since it is sufficient to consider the sign of each signal: the positive signals take a value b = + λ and the negative signals b = -λ, that is:

The dispersion of the phases corresponds to that of a discrete random variable with two values, so the Equations (7) to (9) above apply, substituting φXk) for bXK).

In an even more particular embodiment, the variance of the variable b¡ for a given range / c is:

From this equation it follows that the minimum of σ ² is 0 and, the maximum 1. Logically, also the standard deviation σ has a range of values (0, 1). Thus, in Equation (9), 02 = 1 can be made and the max (-) function eliminated, since the CF factor (k) will take values between 0 and 1. With this, the coherence factor of phase by polarity CFP (k) as:

CFP (k) = (\ - σ ² [bχk) f (12) where the P≥O exponent is a parameter that allows adjusting the correction level. On the other hand, the first term of the numerator of Equation (11) is:

N∑b ^> = N ² (13)

¡= I

Substituting in (11) results:

Substituting in (12):

The variation range of CFP (k) is 0 to 1, for any value of the exponent P. In particular, CFP (k) is zero when the variance is maximum and equal to 1 (see Equation 12), which represents signals with a great diversity of phases and, therefore, of low coherence. Conversely, CFP (k) is unitary when the variance is canceled, a situation that occurs when all signals are in phase and, therefore, are consistent. Thus, the factor CFP (k) can be used to correct the ultrasonic images by application of Equation (4) by substituting CF (TcJ for CFP (k). It should be noted that, from equation (14), σ ² can be cleared) (b¡) and, extracting the square root, obtain the standard deviation σ (b¡) with which an analogous CFP factor (k) is calculated, without the proposed method being substantially modified. The action of the exponent P, which can be programmed by the user, is to emphasize or mitigate the effect of the correction. Thus, for its minimum value P = O, CFP (TcJ = I is independent of the phases of the opening data. In this case, when applying the correction according to Equation (4), the output data equals the input data: y (k) = x (k), that is, no correction is made to obtain the original image.For values 0 <P <1 moderate corrections are obtained, which are accentuated when increasing P. For high P values the effect of The correction increases, reaching a point where only the totally coherent signals are visible, since all the signals have a certain amount of noise, which is not coherent, from a certain value of P they can disappear in the image. true reflectors However, the range of variation of P can be very high, having proven its effectiveness in intervals of 0 to 50.

A second aspect of the invention is directed to an apparatus for the correction of ultrasonic images, which comprises means to calculate a coefficient of phase coherence from the phases of the opening data and a means to multiply said coherence factor by The formed sample.

In a first particular embodiment, the apparatus of the invention uses all the information of the phases of the opening data to calculate the phase coherence coefficient. In that case, the means to calculate the phase coherence coefficient include:

a) A first means of calculation, which receives the signals in phase and quadrature of the opening data, and evaluates the instantaneous phase of the opening data by applying the equation: _{Λ (i) = to} -iWÍ2

^Ψι YES Xk)

The phase and quadrature signals of the opening data are available directly, for example, when a baseband beam shaper is used. In case of not having these signals, for example when the shaper is in radiofrequency, an additional calculation means is necessary connected to the first calculation means, which receives the opening data and calculates the signals in phase and quadrature according to The equation:

SQ ₁ (k) = Hilbert [S ₁ (*)]

SIXk) = SXk)

b) A second means of calculation, connected to the first means of calculation, which determines the phase coherence factors according to the equation:

CF (k) = max (O, l -a-f [φ, (k)])

In a second particular embodiment, the apparatus of the invention takes only the signs of the phases of the opening data. In this second case, the means to calculate the phase coherence coefficient include:

a) An adder, which adds the signs of the phases of the opening data.

b) A table, which receives the output of the adder and the P coefficient, and calculates the phase coherence coefficient according to the equation

In addition, in another preferred embodiment of the invention, the apparatus further comprises a means for detecting equality of signs, which detects the equality of all signs b¡ and sends an indicative signal to the table.

In another preferred embodiment of the invention, the apparatus of the invention comprises means for manually selecting the value of the coefficient P.

Although not explicitly mentioned in the preceding description, the invention also extends to computer programs, particularly computer programs that are located on or within a carrier, adapted to carry out the process of the invention. The program may have the form of source code, object code, an intermediate source of code and object code, for example, as in partially compiled form, or in any other form suitable for use in the implementation of the processes according to the invention . The carrier can be any entity or device capable of supporting the program.

For example, the carrier could include a storage medium, for example, a ROM, a CD ROM or a semiconductor ROM, or a magnetic recording medium, for example, a flexible disk or a hard disk. In addition, the carrier can be a transmissible carrier, for example, an electrical or optical signal that could be transported through electrical or optical cable, by radio or by any other means. When the program is incorporated into a signal that can be directly transported by a cable or other device or means, the carrier can be constituted by said cable or another device or means.

As a variant, the carrier could be an integrated circuit in which the program is included, the integrated circuit being adapted to execute, or to be used in the execution of, the corresponding processes.

DESCRIPTION OF THE DRAWINGS

To complement the description that is being made and in order to help a better understanding of the characteristics of the invention, according to a preferred example of practical implementation thereof, a set of drawings is attached as an integral part of said description. where, for the purposes of illustration and not limitation, the following has been represented:

Fig. 1 shows the typical architecture of a conventional digital beam shaper, highlighting the place of insertion of the image corrector proposed in this invention.

Fig. 2 shows a lateral pattern of the ultrasonic beam, to facilitate the identification of the different lobes involved in the formation of the image.

Fig. 3 shows the principle of obtaining the CF phase coherence factor for baseband and radiofrequency beam shapers. Fig. 4 shows the general circuit to obtain the phase coherence factor by polarity CFP in any beamformer and the possible inclusion of optimizations.

Fig. 5 shows an embodiment example for 32 channels, which calculates the phase coherence factor by CFP polarity in real time, with which it corrects the images obtained by the beam shaper, highlighting the innovation introduced by this invention.

Fig. 6 presents a graph that relates the values of the phase coherence factor by polarity CFP with the number of coherent signals in a 32-channel system, for different values of the exponent P.

PREFERRED EMBODIMENT OF THE INVENTION

Prior art

A system according to the prior art has the structure and devices shown in Figure 1, general architecture that is well known in the field with various variants. The array (10) of transducers is composed of the N numbered transducer elements (1), (2), ..., (N). Each element generates an ultrasonic pulse when excited by an electrical signal and, reciprocally, generates electrical signals by receiving ultrasonic echoes. In emission, the switch (11) connects the array elements (10) to the drivers (19) and, in reception, to the amplifiers (12).

To generate an ultrasonic beam in emission, the N exciters (19) are activated at calculated and coordinated time intervals to produce the deflection and focus of the beam in one direction and depth determined. At the end of the excitation of the elements, the switch (11) goes to the receiving position.

The generated ultrasonic beam propagates through the inspected medium (21), producing echoes in each discontinuity. These echoes return to the array (10) where they are received by the N elements (1) through (N). The signals pass through the switch (11) and are amplified by N amplifiers (12), optionally with different gains to perform the apodization operation. The amplified signals are digitized by N analog-digital converters (13) independently. The outputs R ₁ , R ₂ , ..., R _N of the A / D converters are connected to independent delay devices (14) for each signal. The delays are adjusted to compensate for differences in flight time from the emission to the focus and to each element from the control device (22).

Thus, the data set of the opening (20) is obtained, composed of the N delayed signals S ?, S ₂ , ..., S _N - An adder (15) performs the sum of these signals to obtain the output x, which is focused by the delays applied to the set of signals Ri, R ₂ , ..., R _N - THE most advanced systems dynamically modify these delays, in a beam shaper, to follow the ultrasonic pulse in its propagation through the medium ( 21), so that the signal x obtained is focused on its entire length (dynamic focusing technique).

Once formed, the signal x passes to an envelope detector

(16) and subsequently to a sweep coordinate converter (17), finally being displayed on the screen (18).

This process is repeated for a variety of addresses, changing the delays in emission and reception, so that a region of interest is explored. If the array is linear, the scan occurs in a flat and, if two-dimensional, in one volume.

The effect of the conformation of the ultrasonic beam in emission is that a reflector in the vicinity of the focus produces a high intensity echo which, in turn, is focused on reception by compensating the flight times to each element by the introduction of the corresponding delays.

However, the conformation of the beam is not perfect. Usually, the amplitude of the output of the shaper is described by the lateral pattern of the beam. Figure 2 shows in logarithmic scale (dB), the lateral beam pattern of an array of 64 distanced elements λ and a relative bandwidth of 40%, for a deflection angle θ ₀ = 20 °, position in which is the main lobe (A). The presence of lateral lobes (B) is observed, especially high in the vicinity of the main lobe to which they widen, and a large grid lobe (C). The grid lobe appears when the distance between array elements is greater than half a wavelength (λ / 2), as usual with scattered openings. The lateral lobes next to the main one reduce the lateral resolution of the imaging system.

A reflector located in the direction θo of the main lobe (A) produces a maximum amplitude at the output of the shaper (0 dB) for the focal law that corresponds to the deflection θo and its range. But, when the focal law is modified to visualize the signals coming from the θi direction, where there is no reflector, at the exit of the shaper a signal is obtained with the amplitude corresponding to the grid lobe, due to the replica of the reflector in θo. Similarly, in the directions of the lateral lobes, at the exit of the shaper, the corresponding amplitudes will be obtained, although there are no reflectors. On the other hand, the delays are calculated for a given ultrasound propagation speed, but the variations that it suffers in its propagation by non-homogeneous means are unknown, producing focusing errors or phase aberrations that blur the image.

The level of these false indications limits the dynamic range of the image, as well as the contrast between anechoic zones and areas with reflectors or dispersers, so it is very convenient to have means that reduce the level of the lateral and grid lobes.

In addition, the width of the main lobe and the nearest lateral lobes determines the lateral resolution of the imaging system, that is, its ability to discriminate between two nearby reflectors. Thus, it is also desirable to reduce the width of the main lobe simultaneously with a reduction in the level of the lateral lobes to improve the resolution of the ultrasonic imaging system.

Example 1

Figure 3 shows the block diagram of an apparatus (55a) according to the invention, in which the phase coherence factors are calculated, according to the procedure described above in this document, using all the information of the phase of the opening data.

In an embodiment of the invention, for example when a baseband beam shaper is used, only one block (31) is required to evaluate the instantaneous phase of the opening data by application of Equation (5). In another embodiment, for example when a radio frequency beamformer is used, a Hilbert transformer (30) to carry out the operation of Equation (6).

At the exit of the block (31), therefore, the phases of the opening data are obtained, from which the block (32) determines the values of the phase coherence factors according to the formula:

CF (k) = m _a χ (O, \ - af [φXk)]) (16)

where f [φ _t (k)] is a dispersion estimator, preferably the standard deviation or the variance of the N phases φ _t (k) for each range k, it is already an adjustable constant that determines the sensitivity of the factor CF (k ) to the dispersion of the phases.

Example 2

An example of embodiment of an apparatus (55b) according to the invention is described in the case of using only the signs of the opening data to calculate the phase coherence factors. Figure 4 shows a simple electronic scheme used to implement Equation (15). An adder (40) obtains the output SQ = Zq ,, which is the sum of the sign bits qi, Q2, ..., QN of the data of the opening Si, S2, ..., SN obtained by an apparatus conventional (see Figure 1). It should be noted that the adder (40) interprets the value of the sign q of the signals in complement to 2, that is, q = 0 for the positive signals {b = + ^) and q = 1 for the negative signals (6 = -1) .

The sum SQ of the N signs can only produce values in a set that has Λ // 2 + 1 elements. This is the maximum number of entries needed in the table (41) for each value of P. There are some optimizations that can be made to further reduce the amount of resources used. So, the case | SQ \ = N, which occurs exclusively with an equality of all the signs that _h can be detected separately with the circuit (43), indicated by strokes as its presence is optional. The equality of signs is equivalent to a zero variance, with which CFP = I according to Equation (12). The output of the equal sign detector (43) activates the input U of the table (41) so that it provides a unit value at its output, reducing the total number of inputs required in the table (41) to Λ // 2 . Alternatively, the case in which | SQ | = 0, situation in which CFP = O. In this case, the input U of the table (41) is used to provide a null value at its output, also reducing to N / 2 the total number of entries required in the table (41). For example, in a typical case with Λ / = 128, the table (41) contains 64 entries with the optimizations described.

In a possible embodiment, each change in the value of P loads new values in the table (41). In general, the time invested in this operation can be ignored (writing some tens of data).

In another possible embodiment, P can be encoded to act jointly with SQ as an address in a single table, avoiding reloading of the table (41) (in Figure 4 the input of encoded P is indicated by dashed line). In the previous example, for 16 P values, the total table (41) would contain 16x64 = 1024 entries. The access address to the table (41) consists of two fields: the selector for the exponent P and the selector of the CFP value for the current SQ value. It should be noted that the selector of the exponent P does not have to match the value of the exponent, but is a code assigned to a value not necessarily integer. For example, consecutive selectors 0, 1, 2, 3, etc. can be assigned to values P = 0, 0'5, 1, 1 '5, etc. The CFP value obtained from the table (41) weights in the multiplier (42) the signal x corresponding to the sample formed, to deliver the signal to the output and duly corrected with phase coherence by polarity. The addition of a smoothing filter between the output of the table and the input in the multiplier will allow the elimination of transients in CFP, without this implying a substantial change, which is why it is not indicated in the figure.

Example 3

Figure 5 shows the scheme of a particular embodiment of a corrector (55c) of ultrasonic images by phase coherence of the invention for a system of N = 32 elements. In this case, phase coherence by polarity was used, since its implementation is simpler and, in addition, the exponent P = 1 (equations 15 and 18) was considered.

The inputs to the corrector (55c) S- _\ , S ₂ , ..., S ₃₂ were the signals obtained after applying the focusing delays to the signals received by the N elements of the array. Each S signal is expressed in complement to 2 with 12 bits. The sign is indicated by the most significant bit g ?, g ₂ , --- Q32, which is interpreted by the adder (51) as +1 if the signal is positive and by -1 if it is negative.

To facilitate the interpretation of the process, Figure 5 includes the adder (50), which belongs to the shaper, to which 12-bit inputs arrive, producing the 17-bit x output (sum of 2 ⁵ 12-bit values) . On the other hand, the sign bits qi, q ₂ , ... q∞ were added in the adder (51), which produces the SQ output, expressed in the range (-32, 32). Table (52) was constructed with the entries corresponding to Equation (15) for Λ / = 32 and P = 1 using fractional arithmetic.

Table (52) is a RAM memory to be able to modify its content, in which SQ acts as an address in reading, providing the following CF outputs depending on the value of the possible values of the absolute value | SQ |:

For SQ = O, CF = O, exception that could be detected separately to reduce the number of entries in table (52) from 17 to 16, aspect not optimized in the current example.

The values in the table are between 0 and 1, and can be expressed with fractional arithmetic. In this case, they are expressed with 10 bits.

The CF output of the table (52) (10 bits) is multiplied by (53) by the output x of the coherent sum (17 bits) to obtain the xc signal of 17 + 10 = 27 bits of which only the 17 most significant for operating with fractional arithmetic.

The set of devices (55b) constitutes the phase coherence corrector that must be added to the beam shaper to improve the ultrasonic images in lateral resolution, dynamic range, contrast and signal / noise ratio according to the principles set forth in this invention.

Figure 6 graphically shows the resulting value of CFP in 2 í

SQ function for different values of P and example 3 considered in Figure 5. With a solid line the values corresponding to this example are represented with P = 1; with lines of lines the values corresponding to a case P = 2 are shown and, with dotted line, to the case P = O.5. The graph illustrates the faster reduction of CFP by increasing the value of P, which emphasizes the effect of the correction with a greater reduction of the signals detected as non-coherent.

Claims

1. Procedure for the correction of ultrasonic images by phase analysis, characterized in that it comprises the operation of multiplying the sample formed by a phase coherence factor (CF (k)) based on the dispersion of the phases of the opening data

(SXk)).

2. Procedure for the correction of ultrasonic images according to claim 1, characterized in that the phase coherence factor (CF (Jc)) is calculated according to the expression:

CF (k) = max (O, l - a-f [φχk)])

where f [φχk)] is a function of measuring the dispersion of the phases φXk) of the opening data and a modifiable adjustment parameter.

3. Procedure for the correction of ultrasonic images according to claim 2, characterized in that the function f [φXk)] of measuring the dispersion of the phases φXk) of the opening data is the standard deviation σ.

4. Procedure for the correction of ultrasonic images according to claim 3, characterized in that the coefficient a

takes values belonging to the range π

5. Procedure for correction of ultrasonic images according to claim 2, characterized in that the measurement function f [φχk)] of the dispersion of the phases φ _t (k) of the opening data is variance σ.

6. Procedure for the correction of ultrasonic images according to claim 5, characterized in that the coefficient a

takes values belonging to the range oΛ π

7. Procedure for the correction of ultrasonic images according to claim 1, characterized in that the phase coherence factor CF (Jc) is calculated from the signs of the phases of the opening data (S ₁ (Jc) ).

8. Procedure for the correction of ultrasonic images according to claim 7, characterized in that the phase coherence factor CF (Jc) is calculated according to the expression:

where and P is a modifiable adjustment parameter.

9. Computer program comprising instructions of the program to make a computer carry out the method according to any of claims 1 to 8.

10. Computer program according to claim 9, incorporated into storage media.

11. Computer program according to claim 9, supported on a carrier signal.

12. Apparatus (55a, 55b, 55c) for the correction of ultrasonic images by phase analysis according to the method of any of the preceding claims, characterized in that it comprises:

means (30, 31, 32, 40, 41, 43, 51, 52) to calculate the phase coherence factor from the phases of the opening data; Y

a means (33, 42, 53) to multiply said phase coherence factor by the value of the formed sample.

13. Apparatus (55a) for the correction of ultrasonic images according to claim 12, characterized in that the phase coherence factor is determined using all the information of the phases of the opening data, where the means to calculate the factor Phase coherence include:

a first means of calculation (31), which determines the instantaneous phases of the opening data; Y

A second calculation means (32), connected to the first calculation means (31), which determines the phase coherence factors according to the equation:

CF Xk) = max (0, 1 - af [φ, (£)])

14. Apparatus (55b, 55c) for the correction of ultrasonic images according to claim 12, characterized in that the phase coherence factor is determined using the signs of the phases of the opening data, where the means for calculating the factor Phase coherence include:

an adder (40, 51), which adds the signs b¡ of the phases of the opening data; Y

- a table (41, 52), which receives the output of the adder (40, 51) and a coefficient P, and calculates the phase coherence factor according to the equation:

15. Apparatus (55b) for the correction of ultrasonic images according to claim 14, characterized in that it also comprises a means for detecting equality of signs (43), which detects the equality of all signs b¡ and sends an indicative signal to the table (41).

16. Apparatus (55b) for the correction of ultrasonic images according to claim 14, characterized in that it further comprises means for manually selecting the value of the coefficient P.