Systems and methods for forming ultrasound reflectance images

Abstract

Object image values for each location in the ultrasound reflectance image are calculated by summing image values of images associated with selected intersections of different rows and columns in the ultrasound transducer array. Image values for each of the intersections are determined by transmitting an ultrasound transmission signal from the ultrasound transducer at a first selected first sampling position of sub-samples along a column of the intersections, and receiving reflections at least partially simultaneously from a second selected second sampling position of sub-samples along a row of the intersections. Image values for the different locations are calculated by synthesizing the vertical resolution from the ultrasound subsampled transmit signal and the horizontal resolution from the subsampled receive signal. The range of the sub-sampling positions of each intersection point has a large overlap with the sub-sampling range of an adjacent intersection point, and the sub-sampling positions of different intersection points are at least partially dispersed among one another in the overlap. Extending the sub-sampling positions within a larger overlap range can ensure high resolution of the intersection image, while using different sampling points at the overlap can reduce artifacts in the sum image.

Description

Technical Field

[0001] The present invention relates to ultrasonic imaging using an ultrasonic transducer array. Background Art

[0002] At IEEE IUS2021, van Neer et al. described in a poster presentation a large-area two-dimensional piezoelectric pillar array on a substrate used as an ultrasonic transducer, titled "Flexible Ultrasonic Transducer Array Concept Scalable to Large Areas: The First 128-Element 7MHz Linear Array". At IEEE IUS2021, van Peters et al. described in a poster presentation the technology for manufacturing such arrays, titled "Technology for Large-Area and Flexible Ultrasonic Patches". The array includes small-sized uniform pillars (e.g., 40x40x80 microns), with a small continuous pitch between the pillars (e.g., 60 microns), and the array can be of large size (e.g., 150x150 mm, with the number of pillars exceeding six million). The array can be used to form three-dimensional images or one or more two-dimensional images that are reflected from positions in the object space having a distance in a direction perpendicular to the array. The flexibility of the substrate can be used to make the array flexible. Due to the flexibility of the substrate, the array can be used as a wearable device on the human body. This enables images for medical purposes to be obtained during a patient's normal daily activities.

[0003] However, using a large array requires a relatively long time to perform the measurements and calculations required for imaging. Different from reading data from a memory array or reading an optical image from an optical sensor array, sensing the signals from individual receiving elements in an ultrasonic receiving array must be performed as a function of time. This may require repeating many different signals continuously to collect sufficient information for imaging, and the digitized signals from many individual receiving elements must be combined to obtain three-dimensional high resolution.

[0004] The large amount of time required to obtain an image limits its practicality. For example, when the device is used as a wearable imaging device, it cannot be expected that a person will remain stationary during the measurement. In addition, a large amount of processing time may be required to perform the necessary signal processing. This can be reduced by using more powerful processing circuits, but this may prevent wearable use. By receiving ultrasonic waves from array elements at row positions in the array in parallel and / or transmitting ultrasonic waves from different positions in parallel, the time required for such measurements can be reduced to a certain extent. By using wiring more complex than a simple array, the degree of parallelism can be further increased, but excessive complexity is preferably avoided.

[0005] Another way to reduce the time required to obtain an image is to use spatial subsampling, e.g., by only processing signals received at some of the positions in the array. Arrays with pitches of 50 - 500 microns can produce higher resolutions than are required for many applications (e.g., obtaining large scale medical images), so the resolution loss due to subsampling may not be a problem. However, spatial subsampling may introduce image artifacts (e.g., aliasing), or may require sacrificing resolution to suppress such artifacts. SUMMARY OF THE INVENTION

[0006] One object is to provide efficient ultrasonic imaging using an ultrasonic transducer array having a large number of transducers.

[0007] The method according to claim 1 is for forming an object image of ultrasonic reflectivity in an object space, using an array of rows and columns of ultrasonic transducers to define a sensing surface adjacent to the object space. In the method, an object image value for each position in the image is calculated by summing intersection-based image values for positions determined for a plurality of intersection positions, each intersection position being associated with a respective combination of a row and a column in the array. The intersection-based image values for positions in the image are determined using transmissions from subsampled positions from a series of positions that are no further from the column of the respective intersection position than from the row of the respective intersection position (preferably in the column of the respective intersection position) to provide synthetic directional sensitivity at least in the column direction of the array, and receiving or generating reflections from subsampled positions from a series of positions in the row to provide synthetic directional sensitivity in the row direction of the array. The ranges of at least adjacent intersections overlap with each other, e.g., such that the ranges are two, three, four or more times the distance between intersections. The sampling positions used for different intersections in the overlap are at least partially interleaved with each other.

[0008] The intersection-based image values calculated for different positions in the object space may constitute a complete image as a function of these positions. However, although the intersection images used are of the same (part of the) object space, the intersection-based object images determined for different intersection positions are different because different subsampling is used within the ranges in the row direction and the column direction for different intersection-based images. The subsampling may be very deep, e.g., on average, only one tenth of the positions in each row are used along the row and with a certain offset from that row. Subsampling based on random selection may be used.

[0009] Depth subsampling saves computation time but introduces sampling artifacts. By using the summing of intersection-based image values, combined with different sampling selections for at least partially overlapping ranges of different intersections, these sampling artifacts can be reduced. In this way, the sampling artifacts in different intersection-based images at least partially compensate for each other in the sum image.

[0010] Preferably, the scattered subsampling positions are selected to avoid coincidence between the subsampling positions and, for adjacent intersection positions, coincidence between the subsampling offset positions in a row (e.g., between the subsampling positions in a column). However, even if there is partial coincidence, the subsampling artifacts in different intersection-based images at least partially compensate each other in the sum image. Such coincidence can be easily avoided using depth subsampling.

[0011] The at least partially scattered sampling positions can be achieved by independently randomly selecting the sampling positions at different intersection positions. This may result in coincidence of some sampling positions at different intersection positions, but this will rarely occur using depth subsampling.

[0012] In one embodiment, the first selection and the second selection are non-periodic selections. This can be used to further reduce the visibly apparent subsampling artifacts. In another embodiment, the distance values between consecutive positions of the first sampling positions and / or the distance values between consecutive positions of the second sampling positions are distributed within a predetermined range of distance values that includes a plurality of distance values. Thereby, strong local sampling density variations can be prevented. The distribution of the distance values can be similar to a pseudo-random selection from the predetermined range of distance values. In another embodiment, the lower limit of the predetermined range of distance values is at most half of the average value of the distance values.

[0013] In one embodiment, the intersections are located on a two-dimensional position periodic grid with a period smaller than the first and second ranges. This provides uniform intersection image determination. However, in other embodiments, non-periodically positioned intersections can be used to reduce artifacts.

[0014] In one embodiment, the average subsampling rate of the first and second selections is at most half of the intersection density along the columns and / or rows of the corresponding intersection positions. Thereby, depth subsampling can be used. Generally, this would result in strong artifacts, but using the summation of intersection images based on different selections from overlapping ranges enables the use of such depth subsampling with fewer artifacts.

[0015] In one embodiment, the summation of the intersection-based image values of the positions can be a weighted sum. Thus, for example, intersections that are farther from the object space position for which the image values are calculated for distance can be given less weight or no weight in the intersection image. This can be used to reduce noise.

[0016] In another embodiment, the weighted summation defines a cross - point position window, and the first and / or second ranges of the window all overlap with the first and / or second ranges of at least one cross - point position in the window. Thus, artifact reduction can be optimized. At least one cross - point position in the window can be the cross - point position closest to the object space position where the image value is calculated. Preferably, cross - point positions whose ranges do not overlap with the ranges of at least one cross - point position are not weighted. Thus, noise can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] These and other objects and advantageous aspects will be described in the detailed description of exemplary embodiments with reference to the following drawings.

[0018] Figure 1 A top view of an ultrasonic imaging device is shown.

[0019] Figure 2a An imaging system is shown.

[0020] Figure 2b A signal flow model is shown.

[0021] Figure 2c A matrix with reduced number of elements is shown.

[0022] Figure 3 A linear excitation / reception mode is shown.

[0023] Figure 4 A side view of a wearable ultrasonic imaging device is shown.

[0024] Figure 5 A circuit diagram of an array element is shown.

[0025] Figure 6 An imaging flow chart is shown.

[0026] Figure 7 An image calculation flow chart is shown.

[0027] Figure 8 An embodiment with multi - row reception is shown. DETAILED DESCRIPTION OF THE INVENTION

[0028] Figure 1An example of an ultrasonic imaging device is shown. The device has a two-dimensional array 10 of ultrasonic transducer elements that extends laterally in the x and y directions when viewed from a z-direction position perpendicular to the array 10. Imaging using the ultrasonic imaging device involves ultrasonic transmission and reception by the array. Each ultrasonic transducer element can be an ultrasonic receiver, an ultrasonic transmitter, or both. It should be noted that although the layout of the array may seem similar to that of an optical image sensor, ultrasonic imaging is more complex than optical imaging: in addition to the array being used for both transmission and reception, the signals at each array element must be processed as time-dependent ultrasonic signals.

[0029] The array 10 is located on a substrate, which can be a flexible substrate. The array 10 can be rectangular with side lengths of one hundred millimeters or more than several hundred millimeters. At the same time, the array elements can be provided with a smaller pitch of less than one hundred micrometers (e.g., sixty micrometers), resulting in thousands of ultrasonic transducer elements on the sides of the array and millions of ultrasonic transducer elements in the array.

[0030] The ultrasonic imaging device can be used as a wearable ultrasonic imaging device, for example, worn on a human or animal body to capture 3D images of ultrasonic reflectivity within a certain depth range below the area where the device is worn on the body. The device can be used to obtain images of the entire area, or images of much smaller sub-areas, or both. To obtain an overview image of the entire area, the number of array elements is overly redundant. In principle, depth subsampling (e.g., at a rate of one-tenth in the x and y directions) may be sufficient to obtain an overview image.

[0031] The array 10 includes rows and columns of array elements 100 of uniform size. The ultrasonic imaging device is designed to generate ultrasonic signals from array elements 10 at selectable positions in the array or from selectable groups of these positions, and to output electrical signals excited by ultrasonic reception at selectable positions in the array or from selectable groups of these positions.

[0032] A row selection circuit 12 and first and second column access circuits 14, 16 are located near the array 10 on the substrate. In the illustrated embodiment, the first column access circuit 14 can be used to receive signals from the array elements, while the second column access circuit 16 can be used to provide signals for the array elements to transmit. Although an embodiment with two column access circuits 14, 16 is shown for ease of illustration, it should be noted that one column access circuit combining the functions of the first and second column access circuits 14, 16 is sufficient for signal transmission and reception.

[0033] The array 10 may include conductive row lines 24 coupled to a row selection circuit 12 and conductive column lines 23b coupled to column access circuits 14, 16. The row selection circuit 12 has a row address input 120 and an output coupled to the row lines 24. Each row line 24 connects the row selection circuit 12 to the selection inputs of the array elements in the corresponding row of the array 10. The column access circuits 14, 16 have inputs and outputs respectively coupled to the column lines 23b. The column access circuits 14, 16 have column address inputs 140, 160 and signal inputs 162 and signal outputs 142 for ultrasonic excitation signals and ultrasonic detection signals. Each column line 23b connects the column selection circuits 14, 16 to the signal connections of the array elements in the corresponding column of the array 10.

[0034] Figure 2a An imaging system is shown that includes a processing system 40, a wearable ultrasonic imaging device 42 having an array as shown Figure 1 and a timing circuit 43, a signal generator 44, and a detection circuit 45. The signal generator 44 and the row selection circuit 12 and the second column access circuit 16 together form an excitation circuit configured to emit ultrasonic waves from a selectable first ultrasonic transducer in the array. The detection circuit 45 and the row selection circuit 12 and the first column access circuit 14 together form a receiving circuit configured to receive ultrasonic measurement signals from a selectable second ultrasonic transducer in the array. As can be noted, the excitation circuit and the receiving circuit may share some components.

[0035] The processing system 40 has an output coupled to the address inputs of the wearable ultrasonic imaging device 42 for providing column addresses to the column access circuits of the wearable ultrasonic imaging device 42 and for providing row addresses to the row selection circuit of the wearable ultrasonic imaging device 42.

[0036] In addition, the processing system 40 has a control output coupled to the timing circuit 43. The timing circuit 43 has outputs coupled to the signal generator 44 and the detection circuit 45. The generator 44 has an output coupled to the wearable ultrasonic imaging device 42 for providing a transmit signal to the signal input of the first column access circuit of the wearable ultrasonic imaging device 42. The detection circuit 45 has an input for receiving a received signal from the first column access circuit of the wearable ultrasonic imaging device 42.

[0037] In operation, signal generator 44 generates signals transmitted by array elements in selected rows and columns of the array selected by processing system 40. Detection circuit 45 detects signals received by array elements in selected rows and columns of the array selected by processing system 40. Timing circuit 43 controls the generation time of the transmitted signals and the time reference for detection. Detection circuit 45 may be configured to sample and digitize received signals from the array elements, preferably simultaneously sampling and digitizing received signals from multiple columns of the same row, and the time reference defines the relationship between the sampling time and the generation of the transmitted signals. Detection circuit 45 has an output coupled to processing circuit 40 for providing the sampled digitized signals.

[0038] Although the input of detection circuit 45 is shown as a single line, it should be understood that the line of detection circuit 45 may represent a single signal input for receiving signals from a single selected column line or multiple signal inputs for receiving and detecting signals from multiple selected column lines in parallel. In one embodiment, detection circuit 45 has a multi-line input for receiving received signals from all columns of array 10 in parallel. In another embodiment, detection circuit 45 has a multi-line input for receiving received signals from an optional subset of columns of array 10 in parallel.

[0039] Similarly, although the output of signal generator 44 is shown as a single line that may represent a single signal output for providing signals to respective selected column lines, in some embodiments, the line may represent multiple signal outputs.

[0040] Processing system 40 is programmed to control the operation of the imaging system. When operating under the control of processing system 40, processing system 40 calculates an ultrasonic reflectivity image in the object space, which is the space containing points within a certain distance range in the z direction from the xy plane of array 10, and this space will more generally be referred to as the sensing surface. The reflections may come from within an object located in the object space. When wearable ultrasonic imaging device 42 is worn on a human or animal body, the reflections come from within the body at a position where the device is worn to a certain depth inside the body.

[0041] The processing system 40 can be programmed to compute a three-dimensional image of reflectivity, or a two-dimensional image of the reflectivity of a cross-section of an object, where there is a virtual plane in the object space, or a curved virtual surface in the object space. Positions in the object space will be denoted by "r" = (rx, ry, rz), and positions on the sensing surface will be denoted by p = (px, py), which corresponds to p = (px, py, 0) in terms of the object space coordinates. As used herein, an "image" of reflectivity is a set of image values (or its Fourier transform) that corresponds to the reflectivity at positions in the object space, i.e., the set of image values provides information about the position dependence of the actual reflectivity in the object space, not necessarily the actual reflectivity values or without any distortion: the amplitude scale or the spatial scale may be different from the actual reflectivity, and deviations due to the approximate nature of the image computation are not excluded.

[0042] Signal flow of image calculation

[0043] Figure 2b A signal flow model for determining an image Im(r) based on the reflectivity of an object space is shown. The determination of the image involves transmitting a signal from the array elements 100, receiving the reflected signal at the array elements, and computing the resulting image values Im(r) based on the received signals.

[0044] The excitation circuit is represented by a signal generator E1, a vector response filter V(j, r), and spatial sampling functions Sv(j), Av. The receiving circuit is represented by spatial sampling functions Au, Su(j), a vector response filter U(j, r), and a signal detector D.

[0045] In Figure 2b , Av, Au represent sampling because the array 10 consists of discrete array elements. Sv, Su represent subsampling between these elements, which will be discussed later. Av, Au may be absorbed by Sv and Su. Initially ignoring Sv and Su, the transmitted signal for obtaining Im(r) can be expressed as

[0046] e0 = V(r) * s

[0047] Here s is a signal that controls the timing and can be represented as a time-domain signal or a Fourier transform domain signal (this dependence is implicit), and "*" represents convolution in the time domain or multiplication in the Fourier transform domain. "s" is used to distinguish different reception time delays and can be, for example, a pulse signal. Formally, V(r) is a vector whose vector components are for each transmitting element in the array 10, where each component can represent a time-dependent response function that is applied to s to obtain the signal of the corresponding array element, which depends on the position "r" for determining the image value Im(r). It is also possible to select "s" according to the position "r", but this is not necessary, so if there is a dependence on "r", it is not shown.

[0048] In addition, the determination of Im(r) involves a convolution or product sum and scalar product similar to that of the vector D*U(r) and the measured vector signal vs.

[0049] Im(r) = D(r)*U(r)*vs

[0050] Here, vs is formally a vector, whose components are in principle the components of each array element, representing the received signal in the time domain or the Fourier transform domain. U(r) is a vector of the time-dependent response function. D is an operation applied to the scalar product U(r)*vs to select the distance by differentiating the signal values of different time delays. In the embodiment where s is a pulse signal, D can correspond to selecting the value of U(r)*vs with a predetermined time delay. When the signal e is more complex, D can be correspondingly more complex. "D" may depend on the position "r" where the image value Im(r) is determined. However, if there is a dependence on "r", it will not be shown.

[0051] The reflectance in the object space is related to the received signal vs and the transmitted signal

[0052] vs = R'*e0

[0053] where R' is a matrix with rows and columns, the number of rows and columns being equal to the number of transmitting elements and receiving elements in the array 10 respectively (i.e., not equal to the number of rows or columns of the array 10), and "*" represents convolution in the time domain or multiplication in the Fourier transform domain. The matrix R' can be modeled as

[0054] R' = integral of Tu(r')*R(r' - r")*Tv(r") in the object space for r' and r"

[0055] where r' and r" are positions in the object space, R represents the local reflectance (continuous function) in the object space, and Tv and Tu are vectors representing the propagation of ultrasonic waves from positions on the array and via positions in the object space to positions on the array respectively. Tv and Tu are functions of the position r and the travel time in the object space. Formally, the product of Tu and Tv with R involves convolution (or product in the Fourier space) over the object space. However, when the reflectance R can be assumed to be local and instantaneous, the product with Tv will become a simple product. Or more explicitly

[0056] R' = integral of Tu(r')*R(r')Tv(r') in the object space for r'

[0057] Therefore, taking everything into consideration, the image value Im(r) calculated based on vs is equal to

[0058] Im(r) = D*U(r)*R'*V(r)*s

[0059] In principle, the vectors U(r) and V(r) are chosen such that they together compensate for the effects of Tu and Tv. That is

[0060] Im(r) = D * R(r) * s

[0061] Thus, in principle, the computed image value Im(r) of the image corresponds to the reflectivity R(r) at the position "r" in the object space, and this value can be determined by transmitting e0 and receiving vs. Preferably, U and V at least compensate for the propagation time differences represented by Tu and Tv. Thus, the image value approximates R(r). Optionally, U(r) and V(r) can also compensate for the effect of distance on the signal amplitude.

[0062] In principle, the correspondence between R(r) and Im(r) applies to an infinitely sensing surface with infinitesimal array elements. In practice, both are finite. Thus, Im(r) will correspond to a wide-pass spatial filtering version of R, and some high frequencies will be suppressed due to the finite size of the sensing surface.

[0063] As described above, the vectors U(r) and V(r) are preferably chosen such that they can together at least compensate for the propagation time differences represented by Tu and Tv. Preferably, U and V at least provide delay times that vary in a manner opposite to the propagation times represented by the vectors Tu(r') and Tv(r'). The component Tu(r',p) of the vector Tu(r') corresponds to the propagation from the position r' in the object space to the position p of the receiving element. This is a function of p - r'. The travel time of the ray between r' and p is |p - r'| / c, where "c" is the speed of sound. U(r) can be chosen such that the component U(r,p) of the vector U(r) varies in the manner of -|r - p| / c. In the Fourier transform of the time dependence, this corresponds to the phase factors exp(i f|p - r'| / c) and exp(-i f|r - p| / c), where i 2 = -1, and a similar relationship applies to V(r).

[0064] Approximately, the dependence of the travel time difference (|r - p| - |p - r'|) / c on p varies with p*(e(r) - e(r')) / c, where p*(e(r) - e(r')) is the scalar product of p with the orthogonal projections of the unit vectors e(r) and e(r') in the directions of r and r'. This approximation is applicable when the ranges of the propagation distances |r - p| and |r' - p| are small compared to the object distances |r| and |r'|. In this approximation, the Fourier transforms of the time dependences of the components U(r,p) and V(r,p) of the vectors U(r) and V(r) provide the phase factor exp(i k*p), where k = 2*PI*f e(r) / c and "f" is the time frequency. Similarly, in this approximation, the Fourier transform of the time dependence of Tu and Tv contains the phase factor exp(-i k*p).

[0065] Subsampling

[0066] To reduce the time required for measurement and calculation and the storage space required, subsampling can be used. In terms of U and V, subsampling can be modeled as matrix multiplication [U Su], [Sv V] in vector space with the coefficients of the diagonal matrices Su and Sv equal to 1 or 0. Thus, the image values obtained by emission and reception can be represented by the response matrix RT as

[0067] Im(r) = D(r)*[U(r)Su / <su>*RT*[Sv V(r) / <sv>*e

[0068] Among them, <su>is the average value of Su as a function of position, also known as the sampling rate of Su. Similarly, <su>is the average value of Sv as a function of position.

[0069] Subsampling causes a deviation in the correspondence between the calculated image Im(r) and R, that is, artifacts are generated. As is well known, one-dimensional periodic subsampling causes aliasing, that is, folding of the signal spectrum, resulting in high-frequency components in the non-subsampled image generating near-zero frequency components in the subsampled image. Compared with the image without subsampling, the image Im’(r) obtained by subsampling can be expressed by an error term:

[0070] Im’(r) = Im(r) + errU + errV + errUV

[0071] where

[0072] errU = D * U(r)(Su / <su>The integral of -I)*Tu(r’)R(r’)*s with respect to r’

[0073] errV = D*R(r’)*Tv(r’)*(Sv / <sv>-I) Integral of V(r)*s with respect to r'

[0074] errUV = D*U(r)(Su / <su>I)*R(r’)*Tv(r’)(Sv / <sv>The integral of (-I)*V(r)*s with respect to r'

[0075] In the approximation, the dependence of the travel-time difference varies with p*(e(r)-e(r') / c, and in the Fourier transform of the time dependence, U(r)(Su / <su>-I)*Tu corresponds to (Su / <su>-I) in the spatial Fourier transform, similarly, Tv(r’)(Sv / <sv>-I)*V(r) corresponds to (Sv / <sv>-I), a spatial Fourier transform. Thus, in the case of periodic sampling, an effect similar to one-dimensional aliasing occurs. However, as will be discussed, sampling need not be limited to periodic sampling. In principle, the selection of any position p can be used as a subsampling point, even an image point based on a random selection.

[0076] Cross-point image

[0077] In principle, U and V can be vectors having non-zero response function components equal in number to the number of array elements 100 in the array 10. This is equivalent to using all the array elements 100 to create transmit beams focused on different positions r in the object space and further using all the array elements 100 to create selective reception patterns focused on different positions r.

[0078] The actual implementation of the signal flow model may involve making successive measurements using successively different (groups) of array elements 100 as transmitters and receivers. This may result in an overly long measurement time.

[0079] In one embodiment, selected U and V are used to calculate an image to provide resolution in the x and y directions respectively, using transmit elements with a different row offset on the array 10 and receive elements located in the same row (here, jx and jy will be used to represent the number of selected rows and columns in the matrix 10, and j represents the combination of jx and jy).

[0080] In a preferred embodiment, the transmit elements with a different row offset are located in the same column. This can provide the best results. However, after necessary modifications, the transmit elements with different offsets can be located within a range of columns that are selected such that the distance of the transmit elements with different offsets from the reference column does not exceed the distance from a single row. For example, the distance of the transmit elements with a different row offset from the column where the intersection point is located can be no more than one or two array elements. In another embodiment, the transmit elements with different offsets are arranged along a line perpendicular to the row direction, for example, with an angular deviation from the column direction of no more than 30 degrees. They are said to be arranged along this line if the transmit elements are the nearest adjacent transmit elements in the same row to this line.

[0081] The preferred embodiment will be described by way of example below, where the transmit elements with different offsets are located in the same column. In this case, U and V can be regarded as samples of a single row at jy and a single column at jx respectively, which are samples of more general functions U' and V' that depend on two-dimensional jx and jy. The access structure of the wearable ultrasonic imaging device enables it to receive signals in a single row simultaneously.

[0082] Images obtained using the emitter elements in a single column and the receiver elements in a single row in the array 10 may be associated with "intersections", i.e., combinations of the jy-th row and the jx-th column in the array 10. Such images will be referred to as "intersection images" Im(j,r). However, in principle, the image values of such intersection images Im(j,r) are independent of the intersection "j", and the intersection images Im(j,r) for any intersection "j" can be defined over the entire object space and apply to all positions "r" at any distance from the intersection "j" in that space.

[0083] However, when different subsampling selections of the emitter elements in the jx-th column and the receiver elements in the jy-th row in the array 10 are used for different intersections, the intersection images Im(r,j) for different intersections j will be different from each other, although they approximate the same "true" image. Generally, in a slice map of Im(j,r) at a constant distance from the surface of the array 10, the term errU corresponds to artifacts along the horizontal line, the term errV corresponds to artifacts along the vertical line, and the term errUV corresponds to artifacts along the diagonal line. Since depth subsampling may be used, for example, subsampling at less than one-tenth of the positions on the array 10, the deviation may be quite large.

[0084] In one embodiment, the image value of the sum image Im(r) is calculated as the summation (preferably a weighted sum, weighted by the weight factor w(j)) of the image values of such intersection images Im(j,r).

[0085] Im(r) = summation over j of w(j)Im(j,r)

[0086] The summation over j is limited to the intersection images of selected intersections, such as intersections on a grid with a fixed number of columns and rows between adjacent intersections in the array 10. Different subsampling positions on the array 10 are selected to calculate the image values of different intersection images Im(j,r) at the same position "r" in the object space. Therefore, the different deviations caused by subsampling in different intersection images will be summed, and the deviations of different intersection images may cancel each other out, or at least the deviations of each intersection image will not be the same. This can be used to reduce the net deviation in the sum image compared to the deviation caused by subsampling in the individual intersection images.

[0087] Figure 2c A part of the array 10 is shown, where there are only array elements 100 (only one is marked) at selected positions for calculating the image Im(r) according to the intersection image Im(j,r). It should be noted that the array can extend in all directions from the shown part.

[0088] All array elements in all rows are shown only for selected columns 120 that contain intersections. Similarly, all array elements in all columns are shown only for selected rows 110 that contain intersections. Only the array elements in the selected rows 110 serve as receivers for intersections, and only the array elements in the selected columns 120 serve as receivers for intersections. No array elements are required or present at other locations in array 10. In one embodiment, no such array elements are present in the array. However, it should be noted that in other embodiments, there may be more array elements, for example, at all locations in the array. For example, when intersections are used to calculate an image of a wider area, all array elements within a selected area can be used to calculate a local image.

[0089] The number of array elements between consecutive intersections is only an example. In fact, there can be 10 to 100 array elements between consecutive intersections. All intersection images can be calculated in the same way, except for the selection of sampling positions in array 10 used to calculate different intersection images. In one embodiment, the general vectors U(r) and V(r) used in this calculation are the same for all intersection images, but the selection of components of these vectors is different for each different intersection according to the sampling position selection in array 10 for each intersection. In a preferred embodiment, this selection is first made so that the image values of each intersection image Im(j,r) are obtained by emitting ultrasonic waves only from the positions of the array elements 100 in the jx-th column and using only the array elements in the jy-th row. Secondly, for each intersection image Im(j,r), the positions in that row and column used to calculate the intersection image are selected.

[0090] Each intersection image is based on the emission of array elements at sub-sampling positions within a certain column range in array 10 and the reception of array elements at sub-sampling positions within a certain row range in array 10. The sub-sampling positions include the positions at both ends of the range. The row and column ranges of two intersections 130 and 132 are shown as two-dimensional crosshair windows on array 10, which include horizontal ranges 140a and 142a and vertical ranges 140b and 142b. The ranges of these intersections are so large that they overlap each other. As shown, each range in the crosshair windows 140a, 140b and 142, 142b of intersections 130 and 132 contains that intersection. Preferably, the intersection of the ranges 140a, 140b and 142, 142b of each intersection 130, 132 is at least near intersections 130, 132.

[0091] For example, the array size can be 150x150 mm, with 5 - 20 array elements per mm, and the length range of the array can be 20 - 100 mm.

[0092] Using the intersection image obtained by transmitting and receiving, with its array elements dispersed over a position range wider than the distance between intersections, high resolution can be provided in the intersection image.

[0093] Figure 3 Similarly, the intersection positions 50 along a single horizontal line 52 are shown. Different intersection positions are associated with respective vertical lines 53a - e passing through that intersection position. The image of each intersection position 50 is formed using the following signals: an excitation signal from an array element 100 in a first aperiodic subsampled array element position selection along its vertical line 53a - e, and a received signal from an array element 100 in a second aperiodic subsampled array element position selection along the horizontal line 52.

[0094] For the intersection positions 50 of one of the vertical lines, Figure 3 the horizontal range 54 along the horizontal line 52 between the left - most and right - most positions in the second sampling selection of that intersection position 50 is shown. This horizontal range 54 contains the intersection position 50 and extends beyond the adjacent intersection positions 50 in the row direction. Thus, at least for the nearest adjacent intersections in the row direction, the horizontal ranges of the positions of the received signals for the intersections overlap. The positions in the first sampling selection similarly extend beyond a similar vertical range that contains the intersection position 50 and preferably at least contains the vertical ranges of the nearest adjacent intersections in the column direction.

[0095] For each intersection position 50 labeled "j", an ultrasonic signal time - related signal e0(i,j) is excited at the position labeled jy along the vertical line 53a - e of the intersection position j=(jx,jy). Similarly, for each intersection position j, an ultrasonic signal time - related signal vs(jx,jy) is received, and the image value Im(r,j) of the intersection image is calculated using the selected position labeled jx along the horizontal line 52 of the intersection position j. The image values of the same "r" intersection images of different intersections are summed to form the image value of the sum image.

[0096] In the sum image, the bias caused by subsampling is the sum of the biases errU, errV, and errUV in the intersection image. errU and errV respectively appear as biases along the x - axis and y - axis in the image. errUV appears as a bias on the oblique line in the x - y plane of the image. The expressions of errU and errV in the sum image are the same as those in the intersection image, except that the sum of the sampling functions replaces the sampling functions Su(j) / <Su(j)> and Sv(j) / <Sv(j)> of the intersections: for example

[0097] errU = integral of D * U(r) with respect to r' * sum over j of [w(j) * (Su(j) / <Su(j)> - I)] * Tu(r') * R(r') * s

[0098] When the sum of w(j)Su(j) / <Su(j)> over j is closer to a constant function, i.e., when the sampling function equals 1 at more positions on the array 10, the magnitude of the deviation decreases. When using deep sub - sampling, the deviation of a single intersection image is large because in this case the value of w(j)Su(j) / <Su(j)> is far from a constant function.

[0099] However, as shown by the expression of errU, the terms of w(j)Su(j) / <Su(j)> for different intersection images are added. Therefore, using different sub - sampling selections for different intersections can improve the approximation to the constant function. The more dispersed the sub - sampling positions are, the lower the coincidence degree of the sampling position selections for different intersection images "j", and the smaller the deviation caused by sub - sampling in the sum image.

[0100] (The variance of the sum of (Su(j) / <Su(j)> - 1) over j) can be used as a measure of the approximation to the constant function. For a single intersection image, where Su(j) for the sub - sampling selection of m positions out of n possible sampling positions equals 1, the variance of each position (Su(j) / <Su(j)> - 1) is 1 / <Su(j)> - 1. For N overlapping intersection images with non - coincident sampling points, the variance is (1 / N * <su>-1 (No overlapping sampling points means N * <su>Less than or equal to 1), that is, the variance decreases as the number N of intersection points in the overlapping range increases. It should be noted that in this case, adjacent intersection points in the x and y directions are both counted in N, so a rapidly increasing gain can be achieved through a two-dimensional intersection point array.

[0101] The same applies to the deviation errV. The expression of errUV is more complex, but it has been found that reducing the coincidence degree of Su and Sv in different intersection point images will also reduce errUV in the sum image. For both, the gain is optimal when the direction where the emission element is located is perpendicular to the row direction.

[0102] Therefore, in order to make the net deviation in the sum image less than the deviation in each intersection point image due to subsampling, the ranges associated with adjacent intersection points 130 and 132 on the array 10 should overlap, and the subsampling positions of different intersection points in the overlapping area should be at least partially staggered with each other. When different positions in the first set of positions are located between different consecutive position pairs of the second set of positions, the positions of the first set of positions are said to be staggered with the positions of the second set of positions. This does not exclude that there are multiple first set of positions between some consecutive position pairs of the second set of positions, nor does it exclude that there are no first set of positions between some consecutive position pairs of the second set of positions. When the first set of positions and the second set of positions are dispersed within the same position range, they are usually staggered.

[0103] The range associated with the intersection point 130 may overlap with ranges associated with multiple nearby intersection point positions. For example, it overlaps with windows associated with N nearby intersection points in the x and y directions, such as N = 2 or N = 4. Preferably, the range of the nearest adjacent intersection point pair is at least twice the distance between the intersection points. More preferably, it is three times, four times or more times the reduced resolution. A sliding window can be used, with the same size for each intersection point and the same relative position with respect to the intersection point, although the sampling positions within the window may be different with respect to the intersection point.

[0104] The number of subsampling positions in each range can be one-fifth, or less, such as one-twentieth, of the total number of subsampling positions in that range. Preferably, the subsampling rate is high enough so that on average, at least two subsampling positions are located between each intersection point and each of its nearest adjacent intersection points. Preferably, at least 32 subsampling positions are used in each range. A larger number, such as 128 or 256, can be used in each range.

[0105] When the subsampling positions are dispersed throughout the sampling range, the impact of subsampling on the resolution is limited. The zero-frequency peak width of the absolute value square of the Fourier transform of the sampling functions Su and Sv can be used as a measure of both the resolution and the dispersion of the subsampling positions. Preferably, the widths of Su and Sv are selected to be no more than twice the width without subsampling.

[0106] In one embodiment, sampling positions within the horizontal and vertical ranges 140a, 140b of the window for the intersection point 130 are selected such that no sampling position within the horizontal and vertical ranges 140a, 140b of the window 140 coincides with any sampling position of other intersection points 132, where the horizontal and vertical ranges 140a, 140b of the window (e.g., 142) overlap with the sampling positions of the intersection point 130. This can be easily achieved using a lower subsampling rate. The sampling positions within each range are located at positions staggered from the sampling positions within the overlapping ranges of other intersection points. This reduces the net deviation in the sum image. Depth sampling (e.g., using less than one-tenth of the positions in the non-sampled range) enables such a selection even in the presence of a large number of overlapping windows.

[0107] However, even if some sampling positions in the window 140 coincide with sampling positions in the window with overlapping ranges 142a, 142b, as long as not all sampling positions in the overlapping region coincide, the net deviation in the sum image may be less than the deviation in the intersection point image. That is, at least some of the sampling positions in the window are staggered from the sampling positions in the overlapping window, but another part of the sampling positions in the window may coincide with the sampling positions in the overlapping window. Preferably, at least half of the sampling positions in the window 140 do not coincide with the sampling positions in the overlapping window. Similarly, it is not necessary to use all possible sampling positions within the ranges 140a, 140b and 142a, 142b of at least one window.

[0108] Aperiodic subsampling

[0109] Preferably, the selection is non-periodic, i.e., the selected sampling positions of the intersection point are not at integer multiples of a reference distance. The advantage of non-periodic selection is that in the remaining net deviation in the sum image, the aliasing effect is eliminated, thereby reducing the deviation magnitude at the image position r, where the deviation magnitude is the largest.

[0110] Su- <su>and / or Sv- <sv>The position-dependent Fourier transform can be used to characterize aperiodicity. Subtraction of Su and Sv can avoid a peak at zero frequency, which corresponds to unbiased sampling. In the approximation, the dependence of the travel time difference varies with p*(e(r)-e(r’)) / c, and successive operations U(r)Su Tu result in a bias that corresponds to the convolution of R with this Fourier transform. In the case of periodic subsampling, the magnitude of this Fourier transform has sharp peaks at the sampling function, leading to aliasing. In contrast, in the case of aperiodic subsampling, the magnitude is quasi-continuously distributed over a range of frequencies. A measure Q of the aperiodic nature resulting from the received sampling Su is the value of the highest peak except for the peak at zero frequency, provided that the absolute value squared of the Fourier transform of the sampling function Su is divided by the maximum of the absolute value squared of the Fourier transform of the periodic subsampling function with the same number of subsampling positions (same subsampling rate). Preferably, aperiodic subsampling with Q less than 0.5 is used, more preferably Q less than 0.1. With necessary modifications, a similar criterion can also be used for Sv. Sampling selection meeting such criteria can be easily achieved by deep subsampling: when <su>When it is less than 0.5, almost any randomly selected sampling position can meet the requirements.

[0111] Therefore, using subsamples that are dispersed over a range that is wider than the distance between the intersections can provide high resolution in the intersection image. The measurement time required for using such a wider range is reduced by subsampling, and the artifacts generated are reduced by non-periodic subsampling and summing the intersection images obtained using substantially different subsampling selections.

[0112] In one embodiment, the non-periodic selection can be achieved by using frequency modulation, where the consecutive sampling positions p(m) with index m are selected according to p(m + 1) = p(m) + r(m). Here, the frequency r(m) (i.e., the distance between consecutive sampling points) is selected such that the selected distances are distributed within a predetermined distance range (frequency modulation range), preferably excluding distances below a predetermined distance, so as to avoid consecutive sampling positions within that predetermined distance. The frequency r(m) can be randomly selected according to a predetermined probability distribution within the frequency modulation range. Thus, on average <su>is the mean of the probability distribution. For example, the probability distribution can be at <su>remain constant within a finite range centered around it and be zero in other ranges.

[0113] Even if the sampling selection for each intersection strictly follows a periodicity (p(m + 1) = p(m) + d, where "d" is the sampling period), so that the Q value equals 1, the net deviation in the sum image can be reduced by choosing the sampling positions for calculating the images of different intersections. For example, different p(m) mod d values can be used for different intersection images. However, it has been found that using non-periodic subsampling can better suppress the deviation.

[0114] Although an embodiment has been described in which the intersection images are respectively received and transmitted for calculation along a single row and a single column, it should be noted that in addition to a single row, a third option of using multiple rows can also be used. This can be regarded as involving multiple intersections in the same column and different rows, just using the same sampling position in the column. Different rows in the third option can adopt different selections, preferably such that at least part of the sampling positions are staggered with the sampling positions in different second options and the sampling positions of other intersection images. Alternatively, this can be regarded as a single intersection image that involves other rows in addition to the row that defines its intersection.

[0115] Using the same sampling position in the column for multiple rows instead of different selections in the column for each row may increase the deviation in the sum image, but it is still better than using a single intersection image.

[0116] Cross point

[0117] In the shown embodiment, a regular grid of intersections can be used. For example, consecutive intersections with a spacing of 10 - 100 array elements along rows and columns. When using a regular grid of intersections, the selected row 110 and the selected column 120 form a regular grid composed of horizontal and vertical lines. Although Figure 2c An example is shown where one of four rows and four columns is the selected row 110 and the selected column 120, with intersections located therein respectively, but other (usually larger) distances can also be used between the selected row 110 and the selected column 120. However, a regular grid is not necessary. An irregular grid can be used instead. This can further reduce the deviation caused by subsampling.

[0118] It should be noted that the intersection images are not merely separate image blocks. They overlap at most partially with the intersection image blocks of adjacent intersections, and the intersection image blocks of adjacent intersections are not simply spliced together. In principle, each intersection image defines an image value in the same image space, and the intersection images obtained using significantly different subsampling selections are summed to reduce the subsampling artifacts in the high-resolution image obtained under the condition of shortening the measurement time.

[0119] The weights can be independent of the intersection points. Alternatively, a weight factor w(j) can be used to assign variable weights to different intersection point images, e.g., depending on the position "r" in the object space where the sum image is calculated. In other embodiments, the weights w(j) can be used to assign less weight or no weight to different intersection points as the distance from the image point increases, e.g., as a decreasing function of the distance between the image position r and the intersection point position, or at least a non-increasing function. Alternatively, when calculating the sum image for this image point, the image values of the intersection point images of the intersection points "p" that are far from the image point "r" can simply be ignored. For example, the weights can be non-zero (e.g., constant) only for the intersection points closest to the image position for which the image value is calculated and all other intersection points whose sub-sampling range overlaps with the range of this intersection point. This can be used to limit noise when reducing sub-sampling artifacts.

[0120] The weights can be selected such that their sum over the intersection points is equal to 1 in order to normalize the sum image. However, normalization is not necessary because an unnormalized image is also an image. A useful image can be obtained even if the weights w(j) are omitted.

[0121] For large arrays where the length of the rows and / or columns is not much smaller, or even larger, than the distance from the object position "r", when using weighted summation of the intersection point images to calculate the sum image, the weights of the intersection point images can be reduced as the distance between the intersection points and "r" increases. In principle, by summing the intersection point images representing the same object space, it is easy to explain the fact that the most reliable intersection point images are obtained for the intersection points with the shortest distance to "r".

[0122] Device array implementation

[0123] Figure 4 A side view cross-section of the device is shown. In an exemplary example, each array element 100 contains a portion of the piezoelectric material layer 20, where this portion of the layer 20 in the array element forms a piezoelectric column 20a, and this portion of the layer 20 in other array elements is laterally separated by a space 21, which can contain air or a non-piezoelectric filling material. For example, the height of each piezoelectric column 20a can be 80 microns and the cross-section is a square with a side length of 40 microns. The consecutive array elements 100 can be arranged with a spacing of 60 microns.

[0124] For example, the space 21 can be created by hot embossing starting from a continuous layer using a lithographic pattern stamper. The top of the piezoelectric column 20a is provided with an electrical conductor layer 27, which extends above the array 10. In addition, each array element 100 has its own electrode 23a, which is formed in the electrode layer 22 below the piezoelectric column 20a and is used to apply an electric field between the electrode 23a and the conductor layer 27 through its piezoelectric column 20a.

[0125] The substrate on which the array 10 is located can be a flexible carrier foil 26. Optionally, a packaging layer 28 is provided on the electrical conductor layer 27. The row selection circuit 12 and the column access circuits 14, 16 are located on the flexible carrier foil 26 adjacent to the array 10.

[0126] The row lines 24 can be provided below the electrode layer 22 coupled to the row selection circuit 12. The column lines 23b can be provided in the electrode layer 22 and are coupled to the column access circuits 14, 16.

[0127] Each array element 100 can include an access transistor, whose channel is formed in the transistor layer 25 and is coupled between the electrode 23a of the array element 100 and the column line 23b of the column in which the array element 100 is located in the array 10. The row line 24 of the row in which the array element 100 is located in the array 10 can be connected to the gate of the access transistor of the array element 100 or itself form the gate.

[0128] Figure 5 An example circuit diagram of the array element 100 (represented by a dashed rectangle) is shown, which is coupled to the row line 24 and the column line 23b. The array element 100 includes a piezoelectric column 20a, an electrode 23a, and an access transistor 32. The piezoelectric column 20a is located between the electrode 23a and the electrical conductor layer 27, where the electrical conductor layer 27 can be at ground potential. The channel of the access transistor 32 is connected between the electrode 23a and the column line 23b of the column in which the array element 100 is located in the array 10. The gate of the transistor is connected to the row 24 in which the array element 100 is located in the array 10.

[0129] In some embodiments, all the array elements 100 can be designed and connected such that they act as both ultrasonic transmitters and ultrasonic receivers simultaneously. Alternatively, different array elements 100 can also be designed and connected such that they only achieve one of transmission and reception.

[0130] It should be noted that, at an abstract level, the circuit of the ultrasonic imaging device 42 has certain similarities with a DRAM memory circuit or a pixel-based display control circuit. However, its use is at least somewhat different because time-related signals are used to generate and measure ultrasonic signals.

[0131] Processing

[0132] Figure 6 A flowchart showing an operational embodiment of the imaging system is presented. In the first step 61, the processing system 40 selects a crossover point location (e.g., 130 in Figure 2c ), whose index is j=(jx,jy). In the second step 62, the processing system 40 provides an address to enable the first column access circuit to allow the selection signal from the signal generator 44 to be transmitted to the column line 23b corresponding to the crossover point location 130.

[0133] In the third step 63, the processing system 40 causes the row selection circuit to apply a selection signal to the row line 24 to select the row corresponding to the position of the first selected sub-sampling position in the vertical range 140 for the crossover point location 130. While applying the selection signal, the processing system 40 causes the signal generator 44 to generate an ultrasonic excitation signal according to the e0(t) component corresponding to the selected array element, and this signal is transmitted by the first column access circuit to the column line 23b. Under the control of the row line signal, this excitation signal is selectively transmitted to the piezoelectric columns of the array elements 10 of the selected row and column. For example, the excitation signal can be a pulse modulation signal; the delay corresponding to the selected e0(t) component can be calculated separately.

[0134] Subsequently, in the fourth step 64, the processing system 40 causes the row selection circuit to apply a signal to the row controller of the row line 24 to select the row corresponding to the row where the horizontal range 140 of the selected crossover point 130 is located. In addition, the processing system 40 causes the second column access circuit to allow the selected one or more time-related signals received from the array elements in the column corresponding to the second selected sub-sampling position in the horizontal range 140 of the crossover point location 130 to be transmitted to the detector circuit 45. In response, the detector circuit 45 obtains the digital signal values at a series of time points starting from the emission of the ultrasonic excitation signal. Preferably, the signals from multiple columns are simultaneously transmitted to and processed by the detector circuit 45. In one embodiment, the second column access circuit is configured to transmit at least the signals from the horizontal range of the selected crossover point simultaneously. In another embodiment, the second column access circuit can be configured to transmit the signals from the programmed selected columns corresponding to the sub-sampling positions in the row simultaneously. Still in the fourth step 64, the processing system 40 reads the digital signal from the detector circuit 45 and stores it in a memory (not shown).

[0135] The fourth step 64 can be started immediately after the end of the ultrasonic excitation signal in the third step 63. The fourth step 64 can be continuously executed within the time interval of receiving the measurable reflection of the ultrasonic excitation signal. If the detector circuit 45 can only process the received signals from a single column each time, the third to fourth steps 63, 64 can be repeatedly executed for the consecutive positions in the second selected sub-sampling positions in the horizontal range 140 of the crossover point location 130.

[0136] In the fifth step 65, the processing system 40 tests whether all positions in the first selected subsampled positions from the vertical range 140 for the intersection point position 130 have been used. If not, the processing system 40 selects the next position from the first selection in step 65a and repeats the operation for the next position starting from the second step 62. After all positions in the first selection 52 have been used, the processing system 40 performs the sixth step 66 to test whether the previous steps have been performed for all intersection point positions 130. If not, the processing system 40 selects the next intersection point position 132 in step 66a and repeats the operation for the next intersection point position starting from the second step 62.

[0137] Once it is determined in the sixth step 66 that the previous steps have been performed for all selected intersection point positions, the processing system 40 performs the seventh step 67, where the processing system 40 calculates the image.

[0138] The image calculation in the seventh step actually involves applying U(j,r) and V(j,r) to each image position "r" and applying D to the result to obtain the intersection point image value for each intersection point "j", and then (weighted) summing the intersection point image values obtained for the same image position "r" in the intersection point images of different intersection points j. The effect of V(j,r) formally represents a vector whose components represent different array elements, and these array elements represent the signals emitted from these array elements. The effect of V(j,r) can be synthesized by summing the received signals emitted by the respective array elements. Similarly, the effect of U(j,r) can be synthesized by summing the received signals of the respective array elements. Since V(j,r) does not vary with the x-component of "r" and U(j,r) does not vary with the y-component of "r", the calculation of the intersection point image values for some different positions "r" can be shared to reduce the amount of calculation of the intersection point image values.

[0139] Figure 7 An example of the image calculation available for the seventh step 67 is shown. In this example, the intersection point image related to the position is first calculated and then summed. In the first step 71, for each intersection point position j, the combined signal generated by the emission from the first selected subsampled positions in the vertical range 140 of the intersection point position j is synthesized to the corresponding position in the second selected subsampled positions in the horizontal range 140 of the intersection point position j.

[0140] It should be noted that in the signal processing flow, the excitation signal for measuring the reflection at position r among objects is defined by the response function vector V(j,r), whose vector components correspond to the positions in the array and define time-domain or frequency-domain filtering operations. Only selected components of V are used, and these components correspond to the first subsampling selection of the positions in the array, with different first subsampling selections used for different intersections. Similarly, only the first subsampling selection of the positions in the array is used.

[0141] Therefore, the processing system 40 sums the results of all selected measured reflection signals obtained for that position in the second selection by applying the components of the first direction response function vector V(j,r) to the emissions from different positions that respond to the first selected positions 50 at the intersection positions, thereby synthesizing the composite reflection for each position in the second selection.

[0142] The measured reflection signal and the synthesized reflection signal are time-dependent, and the synthesized reflection signal can be calculated for multiple time points. Alternatively, Fourier transform domain calculations can be used to account for the time dependence, including treating the Fourier transform domain signal as a complex phase factor to compensate for the travel time differences in the beam directions from different positions 50 of the first selection.

[0143] In the second step 72, for each intersection position, the processing system 40 synthesizes the reflection signal for the object space direction at the intersection position based on the synthesized reflection signals at the positions of the second selection.

[0144] It should be noted that in the signal processing flow, the received signal for measuring the reflection at position r in the object space is defined by the response function vector U(j,r), whose vector components correspond to the positions in the array and define time-domain or frequency-domain filtering operations. Only selected components of U are used, and these components correspond to the second subsampling selection of the positions in the array, with different second subsampling selections used for different intersections.

[0145] Therefore, the processing system 40 synthesizes the reflection signal for the object space direction by applying the components of the second direction response function vector U(j,r) to all the synthesized reflection signals at the second selection intersection positions and summing the results. The synthesized reflection signal is time-dependent and can be calculated in the Fourier transform domain.

[0146] Since multiple positions are selected in the same column for the first selection, the first step 71 can provide a combined enhanced direction sensitivity when the direction component along the column direction changes. Similarly, since multiple positions are selected in the same row for the second selection, the second step 72 can provide a combined enhanced direction sensitivity when the direction component along the row direction changes. The first step 71 and the second step compensate for the direction-related propagation time difference, but the combined signal may still depend on the time when the ultrasonic excitation signal is emitted. For example, since the reflectivity of the object space changes with the array distance.

[0147] In a third step 73, for each intersection position, the processing system 40 uses the combined signals of different beam and receiving direction combinations to obtain the reflection measurements within a selected time delay after the ultrasonic signal is excited. This corresponds to the effect of the signal processing operation "D".

[0148] In principle, the image values of a three-dimensional image of the ultrasonic reflection in the object space can be calculated in this way. Optionally, the calculation can be limited to the positions on the surface of the object space. In a fourth step 74, the processing system 40 uses the results of different intersection positions to synthesize the sum image of the reflections. Optionally, the results regarding the conformal surface positions in the third step 73 can be used to obtain the results along a conformal surface that is conformal to the array surface, and the distance between the conformal surface and the array surface is d. Another option is to determine the results of a two-dimensional slice of the object space that laterally extends to the array.

[0149] As described above, Figure 7 the examples are mainly used to illustrate one implementation of the calculation. In fact, different implementations can be used. For example, some or all of the calculations related to time and / or spatial positions can be replaced by calculations in the Fourier transform domain. Another example is that the order of the steps can be changed. For example, the signals in the receiving direction can be combined first, and then the signals in the beam direction can be combined, or the signals with a selected time delay can be selected first, and then the signals in the receiving direction and / or the beam direction can be combined. Another example is that the first and second steps 71, 72 can be executed as part of the steps Figure 6 shown, because the signals required for the intersection positions have been recorded, but are executed before all the signals are recorded. For example, if the second step 72 is executed first, it can be executed before using the signals from all the positions 50 of the first selection, or steps 71, 72 can be executed before using all the intersection positions.

[0150] In one embodiment, the total measurement time can be reduced by receiving responses to the same transmit signal on multiple rows. The transmit time intervals during which the array elements are excited to transmit ultrasonic waves are typically much shorter than the receive time intervals during which the reflections of the ultrasonic waves are received. In most embodiments, the selected rows can be changed multiple times during the receive time interval, enabling time-division multiplexed output of reflections from different rows.

[0151] Thus, as Figure 8 shown, in response to the same transmit signal, signals at different intersections on the same column 82 can be received by receivers on different rows 84. The disadvantage of using receivers on different rows 84 to receive signals at different intersections on the same column 82 is that the same vertical subsampling selection will be used for different intersections, which reduces the suppression effect of subsampling artifacts. However, in many applications, a certain degree of suppression of subsampling artifacts can be tolerated.

[0152] Alternatively, this can be regarded as an embodiment in which receivers in multiple rows 84 are used for the same intersection 80 to obtain partial resolution in the column direction, thereby reducing the number of positions required to provide the same resolution separately in the column direction. Thus, receivers in multiple rows 84 are used for one intersection such that one or more of these rows 84 do not intersect the column 82 at the intersection 80.

[0153] In this embodiment, the positions of the array elements used for reception in different rows 84 are combined with the positions of the array elements used for transmission in different rows in the column 82 to synthesize the resolution in the column direction. In this embodiment, the resolution in the column direction can be described by the product of a transmit function and a receive function, where the transmit function and the receive function are respectively the results of synthesizing the array elements used for transmission and the array elements used for reception.

[0154] As in the single-row embodiment, the response function U(j,r) can be selected to compensate for the travel time difference from the positions of the array elements used for reception (which are now also in different rows) to the position "r" in the object space where the image is formed.< / su> < / su> < / su> < / sv> < / su> < / su> < / su> < / sv> < / sv> < / su> < / su> < / sv> < / su> < / sv> < / su> < / su> < / su> < / sv> < / su>

Claims

1. A method of forming an object image of ultrasonic reflectivity in an object space using an array of rows and columns of ultrasonic transducers defining a sensing surface adjacent to the object space, the method comprising: - The object image value at each position in the image of the ultrasonic reflectivity in the object space is calculated by summing the intersection-based image values of the positions determined for a plurality of intersection positions, each intersection position being associated with a corresponding combination of rows and columns in the array; wherein the intersection-based image value for each corresponding intersection position of the plurality of intersection positions is determined by: · An ultrasonic emission signal is emitted from the ultrasonic transducer at a first sampled position of a first selection of subsampling, the distance of the first sampled position of the first selection from the column of the corresponding intersection position not exceeding the distance from the row of the corresponding intersection position, and the first selection extending within a first offset range from the row of the first sampled position; · The reflection of the ultrasonic emission signal from the object space is received at a second sampled position of a second selection of subsampling along the row of the corresponding intersection position, or at a second sampled position of a second selection in a third selection of subsampling including the row of the corresponding intersection position, one or more of the second selections extending within a second range from a second position along the row of the corresponding intersection position, and at least a portion of the reflections from the second selection or the plurality of second selections are received simultaneously; · For the corresponding intersection position, an intersection-based image value associated with the reflectivity at the position in the object space is calculated, the image value being synthesized from the reflections received at one or more of the second selections in response to the ultrasonic emission signal at the first selection; wherein: - The first range of each corresponding intersection position has a first overlap with the first ranges of more than one of the plurality of intersection positions, at least a portion of the first positions of the first ranges of more than one of the plurality of intersection positions being at least partially scattered between the first positions in the first range of the corresponding intersection position, and - The one or more second ranges of the corresponding intersection position have a second overlap with the second ranges of more than one of the plurality of intersection positions, at least a portion of the second positions of the second ranges of more than one of the plurality of intersection positions being at least partially scattered between the second positions in the second range of the corresponding intersection position.

2. The method according to claim 1, wherein, The first range of the corresponding intersection position extends at least to the offset of the nearest intersection position from the corresponding intersection position, and / or the second range of the corresponding intersection position extends at least to the nearest row of the corresponding intersection position.

3. The method according to any one of the preceding claims, wherein, Both the first selection and the second selection are non-periodic selections.

4. The method according to claim 3, wherein, The distance values between consecutive first sampled positions and / or the distance values between consecutive second sampled positions are distributed within a predetermined range of distance values containing a plurality of distance values.

5. The method according to claim 4, wherein, The lower limit of the predetermined range of distance values is at most half of the average value of the distance values.

6. The method according to any one of the preceding claims, wherein, The intersections are located on a two-dimensional position periodic grid, the period of which is less than the first range and the second range.

7. The method according to any one of the preceding claims, wherein, The average subsampling rate of the first selection and the second selection is at most half of the intersection density along the column and / or row of the corresponding intersection position.

8. The method according to any one of the preceding claims, wherein, The summation of the intersection-based image values for the positions is a weighted summation.

9. The method according to any one of the preceding claims, wherein, The window that weights and sums to define the intersection position, where all of the first range and / or the second range of the window overlap with the first range and / or the second range of at least one intersection position in the window.

10. A computer program product comprising an instruction program for a programmable processing system, which when executed by the programmable processing system causes the programmable processing system to perform the method according to any one of the preceding claims.

11. An ultrasonic imaging system, comprising: - An array of rows and columns of ultrasonic transducers that define a sensing surface; - An excitation circuit configured to cause an optional first ultrasonic transducer in the array to emit ultrasonic waves; - A receiving circuit configured to receive ultrasonic measurement signals from an optional second ultrasonic transducer in the array; - A processing system configured to - calculate an object image value for each position in an image of ultrasonic reflectivity in the object space by summing intersection-based image values of positions determined for a plurality of intersection positions, each intersection position being associated with a corresponding combination of rows and columns in the array; wherein the processing system is configured to determine an intersection-based image value for each corresponding intersection position among the plurality of intersection positions by: · causing the ultrasonic transducer to emit an ultrasonic emission signal at a first selected first sampling position of sub-sampling, the first selected first sampling position being no more distant from the column of the corresponding intersection position than from the row of the corresponding intersection position, and the first selection extending within a first offset range from the row of the corresponding intersection position; · receiving a reflection of the ultrasonic emission signal from the object space at a second selected second sampling position of sub-sampling along the row of the corresponding intersection position, or at a second selected second sampling position of a third selection of sub-sampling that includes the row of the corresponding intersection position, one or more of the second selections extending within a second range of a second position along the row of the corresponding intersection position, and at least a portion of the reflection of one or more of the second selections being received simultaneously; · calculating, for the corresponding intersection position, an intersection-based image value associated with the reflectivity at a position in the object space, the image value being synthesized from reflections received at one or more of the second selections in response to the ultrasonic emission signal at the first selection; wherein: - The first range of each corresponding intersection position has a first overlap with the first ranges of more than one of the plurality of intersection positions, and at least a portion of the first positions of the first ranges of more than one of the plurality of intersection positions are at least partially scattered among the first positions in the first range of the corresponding intersection position, and - One or more second ranges of the corresponding intersection position have a second overlap with the second ranges of more than one of the plurality of intersection positions, and at least a portion of the second positions of the second ranges of more than one of the plurality of intersection positions are at least partially scattered among the second positions in the second range of the corresponding intersection position.

12. The ultrasonic imaging system according to claim 11, wherein, - The excitation circuit includes a signal generator and a first access circuit coupled between the signal generator and the columns of the array to generate ultrasonic waves having a time dependence defined by the signal generator from a first optional position in the array, the first access circuit being configured to select the position of the first ultrasonic transducer under the control of the processing circuit; - The receiving circuit includes a detection circuit and a second access circuit coupled between the columns of the array and the detection circuit. The detection circuit is configured to detect the time dependence of an ultrasonic measurement from a selected second ultrasonic transducer in the array, and the second access circuit is configured to select the position of the second ultrasonic transducer under the control of the processing circuit.

13. The ultrasonic imaging system according to claim 11 or 12, wherein, The first range and / or the second range of the respective intersection positions extend at least along the columns and / or rows of the respective intersection positions to the intersection position closest to the respective intersection position.

14. The ultrasonic imaging system according to any one of claims 11 to 13, wherein, Both the first selection and the second selection are non-periodic selections.

15. The ultrasonic imaging system according to any one of claims 11 to 14, wherein, The summation of the intersection-based image values for the positions is a weighted summation.

Images