WO2004055941A1 - Device for generating delays for beam synthesis apparatus - Google Patents

Abstract

The invention concerns a device which consists in storing in three tables (5, 6, 7), for each of the transducer elements of an ultrasound probe, reception time reciprocal values (1/t-tau∞.), offset values (tau∞) and corresponding indexes (ND), and values (Delta<2>) which are a function of the square of the distance of the element K concerned to the central reference element, and in calculating the delay tauK(t) to be applied at time t to said element in accordance with the formula (I).

Description

 DELAY PRODUCTION DEVICE FOR APPARATUS

BEAM SYNTHESIS

The present invention relates to a device for producing delays for a beam synthesis apparatus.

In digital ultrasound devices, the acoustic probe comprises a network of piezoelectric transducers which, in the emission phase, receive electrical pulses and emit ultrasonic beams which they focus at a greater or lesser distance and at an angle d variable incidence depending on the phase shifts or delays of the voltages applied to the various transducer elements. In the reception phase, the acoustic echoes due to the reflection of each incident beam in the body to be explored are picked up by the probe, converted into electrical signals and processed to form an image of the interior of the body being explored. The elementary signals corresponding to the received echoes are delayed according to determined variable delays and added together to form beams, so that these beams correspond to the reception of ultrasonic energy reflected by a single focal point of the body explored.

To create the delays allowing this dynamic focusing to be achieved, lines with variable delay are generally used. For example, US-A-5,844,139 describes a dynamic variable delay production method for an ultrasonic beam former, and discusses the minimization of the cost of a variable digital interpolator, but the problem of cost delay production is not addressed. Another document, the article entitled: "Low Power Delay Calculation for Digital Beamforming in Handheld Ultrasound Systems" by HT. FeldKàmper, R. Schwann, V. Gierenz and TG Noll, published in "2000 IEEE Ultrasonics Symposium", mentions the cost of generating delays and proposes an algorithm to produce these delays, but does not deal with the problem of the cost of computation and the cost of storage, which are far from negligible.

For high-end ultrasound systems, the problem of the cost of delay generation for each element of the probe and for each pointing direction of the emitted beam is relatively negligible, while it becomes critical for portable and / or low-cost devices. . The direct storage of a complete set of delay values for all the sampling periods requires a large memory capacity, but does not require any calculation and can accept any law of evolution of the delay (which may involve an arbitrary anisotropy of the medium in which the ultrasound is propagated).

To save memory space, one could use the incremental representation of the laws of evolution of these delays, because the variation of the delay from one signal sample to the next is very small over a large area, which means that 'usually only one bit is needed to code this variation. However, such a process would require a complex decoding, without however losing either precision or the definition of the general law for the evolution of delays.

If one tolerates a small loss of precision, but no modification of the general law, the laws of evolution of the delays can be approximated by an approximation of function of the first order, like for example a linear function by pieces. The reduction in the memory space required can be very significant, but the computing power required is correspondingly increased.

The subject of the present invention is a method for producing delays for an apparatus for synthesizing transmission and reception beams, these beams possibly being of variable orientation or not, a method which requires neither a large memory capacity nor a large power of calculation, and which is simple to implement.

The method of the invention is a method of producing delays for the various transducer elements of a beam synthesis apparatus, and it is characterized by the fact that the values of inverses are stored in at least one first table. of reception times expressed in multiples of sampling quanta, which are stored in a second table of offset values which are a function of the transducer element considered with respect to a reference element, as well as of the indices relating to the different elements considered, that we store in a third table values of squares of the ratio between the distance of the transducer element considered to the reference element and the speed of propagation of the beams in their propagation medium, and that for each element transducer we subtract the offset value corresponding the product of the value read in the first table by the value read in the third table from which the square of the corresponding offset is subtracted.

The present invention will be better understood on reading the detailed description of an embodiment, taken by way of nonlimiting example and illustrated by the appended drawing, in which:

• Figure 1 is a diagram showing the geometric aspects of the emission of ultrasonic waves emitted by a probe;

• Figure 2 is a diagram of the law of evolution of the delay to be applied to transducer elements as a function of time;

• Figure 3 is a block diagram of a device for producing delays according to the invention.

The present invention is described below with reference to a medical ultrasound ultrasound system, but it is understood that the invention is not limited to this application and that it can be implemented for other training devices. beams, such as sonars, radars

With reference to FIG. 1, we will explain the main geometric relationships governing the relative arrival times of sound waves on the piezoelectric elements of a probe 1. In the drawing, the front face 2 of the probe 1 (face of transmission and reception of ultrasonic waves) has been shown semicircular, but it is understood that the shape of this face can be different (plane, polygonal, ...) without this affecting the geometric relationships developed below .

We neglect here the thickness of the production layer of the face 2, generally made of plastic, and a thickness of about 1 mm. The axis of symmetry of the probe 1 is referenced 3, and it intersects the face 2 at a point C ₀ , behind which is placed the central piezoelectric element (not shown) of the probe. Because, as explained above, the thickness of the protective layer is neglected, it is assumed that the point C ₀ coincides with the center of the transmission / reception face of said central element of the probe. Or a point C | <on face 2, corresponding to the center of the transmission / reception face of any piezoelectric element of rank K, other than the central element of the probe. We call D ^ the distance between C ₀ and C. We assume that at the instant considered the emission and reception of an ultrasonic wave are in a direction referenced 4. Let M be the focal point of the beam emitted in direction 4, M being located on this direction. R _{0 is} the focal distance from the central transducer element

(R ₀ : distance between C ₀ and M), and R the distance between C and M. We call

_{Κ κ} the angle between the half-lines C ₀ C and C ₀ M, and this angle is fixed for a given probe and for a given angle of incidence.

The calculations set out below assume that the ultrasonic waves propagate in a homogeneous and isotropic medium. On reception, the echo beam is focused by applying a delay of its own to each of the transducer elements, in order to align the signals coming from the focal point. It is then possible, thanks to a coherent summation (sum of received signals put back in phase), to amplify the signals as if they came from this focus. The focal point of this echo beam is a point on the body explored and having received the signal emitted. During the observation period, this focal point moves along a line, which is determined by the direction of emission, at the speed (c) of sound in the explored body.

In the diagram in Figure 1, the value of R ^ is given by:

Rκ = C _K MC ₀ M - C _o C _κ = ΛIR: + D - 2R ₀ D _κ cos (Θ _κ )

Compared to the reference transducer (located at C ₀ ), the delay τ _κ (t) to be applied at a given time to the transducer considered (located at C _κ ), is given by:

_ ^R Q - ^R K τ _κ (t)

The instant t at which this delay must be applied is the instant of reception of the echo by the sensor considered, and it is given by:

t _ ^R o ^{+ R} κ In practice, we need to know the relation connecting the delay τ _κ (t) to the instant t of reception by the sensor considered. This relationship is determined as follows:

In the relation: c ² τ _κ (t) “t = R§ -R = 2R ₀ D _κ cos (Θ _κ ) -D ^

because: c (τ _κ (t) + 1) = 2R ^v o ₍ >

we get: c ² • τ _κ (t) • t = c (τ _κ (t) + 1) D _κ cos (Θ _κ ) - D ²

and we get the value of τ _κ (t):

The relation below can also be written

(2)

_ D _κ cos (Θ _κ ) with τ _∞ and Δ = 5κ C cx _∞ being the asymptote of the evolution curve of τ _κ (t) (see in figure 2 the asymptote AS), and Δ ² being a limited set of values which can be stored in a low capacity memory, since D ′ is a function of the number of transducer elements of the probe and c is considered in the present case as a constant for the human body.

Thus, for a given probe and for a given angle of incidence, the law of evolution of the delay τ _κ (t) is of the form 1 / t, to an offset and to a scale factor. An example of this law of evolution has been represented in FIG. 2. In this FIG. 2, the ordinates are graduated in quantum of delay, that is to say, typically, half or a quarter of the sampling period , and the abscissa in sampling periods, and are proportional to the depth of the focal point from which the echo leaves (distance R ₀ ). The formulas set out above make it possible to calculate the delay to be applied to the signal received by each transducer element of a probe with an array of transducer elements so as to dynamically focus the received beam, in order to obtain the maximum spatial resolution. However, this analytical result is difficult to exploit in practice, and the invention proposes to simplify it in order to facilitate its implementation.

Firstly, because the processing of the received echoes is carried out after their conversion into digital form, there is no need to apply the delays except to the sampling rate of the signals received. In practice, it has been found that these delays can be rounded to the value closest to a multiple of the sampling period to obtain acceptable lateral lobes during the formation of the beams, which defines said “quantum of late ”. For example, it may suffice to produce a delay whose value differs by half a sampling period from the theoretical value to obtain lateral lobes meeting the specifications.

Secondly, it will be noted that the delay varies, as a function of time, much more quickly near the probe than far from it. This variation is all the greater as the opening of the area of the body swept by a beam is large. In practice, a minimum exploration depth and a maximum opening near the probe must be taken into account. This is called dynamic opening. Because of these characteristics, the slope of the delay evolution law, that is to say the delay / time ratio, is limited, and the time range, in which the variations of the delay fall within acceptable limits. of the delay quantum, is much larger than the signal sampling period.

Finally, it should be noted that what the digital processing device must produce is not a single delay, but a sequence of delays, in the simplest possible way.

Taking into account these remarks, the method of the invention allows an optimized implementation, and uses three tables and two operators.

The first table used is a table of inverse values, which makes it possible to avoid the division operation in the formula (2) of the delay exposed above. The basic concept of the implementation device is a direct transposition of the result of the analytical calculation described above. he consists of a table of values of the function 1 / T (with T = t - τ _∞ ) of appropriate length and width (i.e. numerical precision and extent). This table is universal, only its addressing and its scale factor are specific to the transducer used and to the speed of propagation of the ultrasonic waves in the body explored.

Each time this table of values of 1 / T is addressed, and the calculation of the delay is carried out, the value obtained remains valid until the next addressing of the table (which corresponds to the address immediately following in this table). Since the error on the determination of the delay must remain less than a quantum of delay, it is generally sufficient to sample this table at a rate lower than the sampling rate, for example once every eight signal samples received .

As specified above, the table of values of 1 / T is independent both of the transducers and of the direction of emission, but it is necessary to access it once by different transducer and once by different direction (in if you have several reception directions simultaneously for the same shot). As a result, depending on the available computing power, it may be necessary to duplicate the table of values by 1 / T so that it is accessible simultaneously to several computing processes. The length of this table is determined by the maximum focusing depth. Typically, 1024 values of 1 / T are sufficient.

To calculate the final value of the delay, you must first scale the values of 1 / T from table 5 by multiplying them by (Δ ² - τ ² j and subtracting them from τ _∞ . The width of this table 5 is the number of bits of the words (numbers) stored in the table. Since the format of these words is fixed, these bits correspond to both the dynamics and the precision (integer and fractional parts) of the words The minimum value of T corresponds to the largest number in the table and therefore to the minimum exploration depth. We found that in most cases, it was enough to define the words of table 1 / T on 12 bits .

We will now describe the device for calculating τ _κ (t) with reference to FIG. 3.

The table of values of 1 / T contains N values, for example N = 1024. It will be noted that among these N values, the values 1 / Tmin and 1 / Tmax are repeated at the beginning and at the end of the table. These values 1 / Tmin and 1 / Tmax correspond to the maximum and minimum exploration depths. The fact that they are each repeated several times (for example 10. times, in a typical manner) is due to the need to display images at the very start of the shooting (in contact with the probe), even if their likelihood is subject questionable because of the spatial selectivity of the piezoelectric elements of the probe because of their non-zero length. Thus, the minimum effective focal length (R ₀ ) is limited to approximately 5 mm, even if observed at a distance of 0 mm, that is to say in contact with the probe. Similarly, maximum exploration is artificially limited, although the display continues beyond this maximum focal length with a slightly defocused beam. This can for example be the case with a linear probe displaying images up to 130 mm with a beam focused at 120 mm from the probe.

The device of the invention comprises a second table 6 of offsets (τ _∞ ). This table contains words made up of two parts of independent types. The first part corresponds to the time offset τ _œ , while the second part is the data ND, corresponds to the offset of the transducer elements relative to the reference (central) element, which is an address index of the third table 7. The information contained in table 6 is only called once at the start of each line of fire, in order to prepare the variables necessary for calculating delays. This table is therefore only called N1 times per firing line, N1 being the number of transducer elements of the probe.

Table 7 contains the values corresponding to the distances between the reference element and the various elements of the probe. In fact, these values are the values of Δ ² (Δ = Dκ / c). These Δ values are rounded and defined, for example on 12 bits.

The devices of the invention also comprise two operators 8, 9. The operator 8 has two main roles: It determines the instant at which the calculated delay must be applied. This is important because the delay values are updated at a lower frequency than the sampling of the echo signal received by the probe. It is therefore necessary to be able to have a reliable initial delay value for minimize the calculation error. This function is called

"Fine timing adjustment".

^ It must provide in table 5 the address of the element for which the delay will be calculated. This function includes the processing of the initial and final periods during which the overflow of addressing by lower and higher values can occur, i.e. the values of t- τ _∞ can be either negative or exceed the capacity of Table. This function is called "rough timing adjustment".

The operator 8 receives as clock signals the same as those for the sampling of the echoes received, and it receives from the table 6 the values of τ _∞ .

The operator 9 supplies a delay line (not shown) connected to the output of the device in FIG. 3 with the delay value to be applied to each current sample of the received signal. This delay is rounded to the nearest delay quantum, and is calculated only once every N1 incident samples, with N1 = 8 for example. This operator calculates the delay according to the relationship:

£ -τl τ _κ t) = τ - t ~ τ „

pour obtenir finalement τ_κ(t) , le termeThis calculation is carried out in two stages: calculation of t ² = Δ ² - τ ² , then calculation of τ _κ (t). The term Δ ² is provided by table 7, while τ ² is supplied by a subset 10 of the operator 9 from the values of t _∞ in table 6, another subset 11 of the operator 9 performing the operation Δ ² -τ ² . Thus, the operator 9 receives τ _∞ from table 6, and he subtracts from it the product (Mt-τ _∞ ) "

to finally get τ _κ (t), the term

(1 / t - τ _∞ ) being provided by table 5 and t ² by operator 11.

Tests have shown that the error on the value of τ _κ (t) was very small: the ratio τ _κ (t) / quantum of delay was, at most, around ± 0.5 quantum.

In what has been explained above, the use of a single table 5 has been described, but it is understood that if we wanted to improve the process, in particular in the case of an environment explored with several different layers, each of them being homogeneous in its volume, it would be possible to use several such tables (or several different zones of the same table), each of them being adapted to each of these layers.

I9ι

Claims

1. Method for producing delays for the different transducer elements of a beam synthesis apparatus, characterized in that the values of reception time inverses expressed in sampling quanta multiples (MX- τ _∞ ), which are stored in a second table (6) offset values (τ _∞ ) which are a function of the transducer element considered with respect to a reference element, thus as indexes (ND) relating to the various elements considered, αue is stored in a third table (7) values of squares (Δ ² ) of the ratio (Dκ / c), between the distance of the transducer element considered to the reference element and the speed of propagation of the beams in their medium of propagation, and that for each transducer element we subtract from the corresponding offset value (τ _∞ ) the product of the value read in the first table by the value read in the third table (Δ ² j from which we subtract the square of the corresponding offset [τ ² ]. 2. Method according to claim 1, characterized in that the minimum and maximum values stored in the first table are repeated. 3. Method according to one of claims 1 or 2, characterized in that the first table is called at a frequency lower than the sampling frequency of the received signal. 4. Method according to one of the preceding claims, characterized in that it is applied to a medical ultrasound probe.