Synthetic data collection method for full matrix capture using an ultrasound array

Abstract

A method for efficiently achieving full-matrix ultrasonic data capture which includes the steps of providing an ultrasound array apparatus, the ultrasound array apparatus further comprising a probe, collecting data over a plurality of inspection locations, generating a plurality of data matrices, each of the data matrices reflecting data collected at the locations, and collecting, initially, a subset of a quantity of data needed for reconstruction of each of the inspection locations. In the method, as the probe moves from collection location to collection location, a data matrix at a prior collection location is gradually filled in as the probe moves to subsequent collection locations. In certain embodiments physical scanning of a probe with few elements is replaced by electronically scanning using an array with many elements.

Claims

1. A method for efficiently achieving full-matrix ultrasonic data capture comprising: (a) providing a two-dimensional ultrasound array apparatus, the ultrasound array apparatus comprising a probe, the probe adapted for scanning and the probe comprising scan increments and nominal scan boundaries, wherein the two-dimensional ultrasound array apparatus has two array dimensions, an array dimension m along a long edge and an array dimension n along a short edge; (b) moving the probe to a first location on the test piece; (c) firing a first element in the long edge of the two-dimensional array and receiving data at all other elements of the array; (d) subsequently firing at each of the remaining further elements along the long edge of array in turn, and receiving on all elements along the short edge of the array; (e) moving the probe over a plurality of inspection locations in a raster scan; and (f) repeating the steps of firing at a first element in the long edge of the two-dimensional array and receiving data at all other elements of the array, and subsequently firing at each of the remaining further elements along the long edge of array in turn, and receiving on all elements along the short edge of the array at the new inspection location.

2. The method of claim 1, wherein the array is operated in pulse-echo mode.

3. The method of claim 1, wherein the probe is scanned a predetermined distance beyond the nominal scan boundaries and the quantity of data is collected such that data matrices of the plurality of data matrices near the nominal scan boundaries are filled.

4. The method of claim 3, wherein each of the dimensions has a number of elements along a length thereof, and wherein the distance the probe is scanned beyond the nominal scan boundaries is equal to the number of elements along each respective dimension minus one.

5. The method of claim 4, wherein there are k scan points along the n dimension and 1 index, or step, points along the m dimension.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a composite schematic illustration of how an array probe operating according to the present efficient data collection method compares with standard data collection;

(2) FIG. 2 is an illustration of a two-step process specific to a pitch-catch probe arrangement where at each collection location, in addition to the transmit element firing, the opposing element from the catch probe also transmits a signal and all elements from the transmit probe (save for the one which originally transmitted) collect the receive signals; and

(3) FIG. 3 is an illustration of a configuration for a two-dimensional ultrasonic array, including a representation of array indices as well as orientation with respect to scan and index axes.

DESCRIPTION OF THE EMBODIMENTS

(4) Referring now to the drawings, FIG. 1 illustrates, in the section identified as Efficient Data Collection, the present, novel ultrasound data collection method. Following the Standard Data Collection method, in order to collect all data for the three probe positions a total of 9 element firings would be needed and 18 waveforms must be collected. According to the present Efficient Data Collection method only 5 element firings and 12 waveforms need to be collected in order to have all waveform data at all positions. In order for this particular implementation to be possible the probe increment must be equal to element pitch. Also, this particular implementation relies on the reciprocity principle (i.e. a signal transmitted by element x and received by element y is in principle the same as a signal transmitted by element y and received by element x). In the case of pulse-echo operation the consequence of the reciprocity principle is that the matrix is symmetrical about its diagonal; i.e. element xy is equal to element yx. At each probe location only one element is fired and all elements receive. As a result, a subset of the data needed for reconstruction at that location is collected. As the probe moves from position A to B to C, the data matrix at location A is gradually filled in. While only three firings are shown, as the probe continues to move the data matrices for locations B and C will be filled as well. Considering, for example, the data matrix at location A: as the probe moves, data from locations A, B, and C are all used to enable reconstruction at that point.

(5) The advantage in efficiency provided by embodiments of the present invention increases as the number of elements and positions increases. For k positions along an inspection line and m elements, number of firings needed and waveforms to be collected are expressed below as

(6) $\mathrm{Number}\mathrm{of}\mathrm{firings}\left(\mathrm{pulse}-\mathrm{echo}\right)=\left(k+m-1\right),\mathrm{and}$$\mathrm{Number}\mathrm{of}\mathrm{waveforms}\mathrm{needed}\left(\mathrm{pulse}-\mathrm{echo}\right)=\left(k.Math.m\right)+\frac{m.Math.\left(m-1\right)}{2}.$
So, in the situation in which a 32-element array is used and 100 data points are taken along a scan line, only 3,696 waveforms will need to be collected and only 131 element firings will be needed. This compares very favorably to the (3233/2)100=52,800 waveforms that would need to be collected from 3200 firings in the absence of the presently-claimed invention, which takes advantage of data redundancy.

(7) If the probe increment is set to a unit fraction of element pitch then the procedure outlined in paragraph 20 may still be applied. In the case that probe increment is equal to element pitch divided by L, then conceptually L arrays may be created and data will be collected for each one in turn every L scan increments. In order to fill all data matrices completely the number k of inspection locations must be evenly divisible by L. Then, for a total of k positions the number of firings necessary and number of waveforms needed are expressed in the equations below as:

(8) $\mathrm{Number}\mathrm{of}\mathrm{firings}\left(\mathrm{increment}\mathrm{is}\mathrm{unit}\mathrm{fraction}\mathrm{of}\mathrm{pitch}\right)=k+L.Math.\left(m-1\right),\mathrm{and}$$\mathrm{Number}\mathrm{of}\mathrm{waveforms}\left(\mathrm{incement}\mathrm{is}\mathrm{unit}\mathrm{fraction}\mathrm{of}\mathrm{pitch}\right)=\left(k.Math.m\right)+L.Math.\frac{m.Math.\left(m-1\right)}{2},$ where k>L for both of the above equations.

(9) If the inspection is performed using a pitch-catch arrangement then a second step is required in order to complete the subset of data that must be collected at each location. Elements normally arranged as receivers must be adapted as transmitters as well, and conversely elements arranged as transmitters must be arranged as receivers as well. At each collection location, in addition to the transmit element firing the opposing element from the catch probe must also transmit a signal and all elements from the transmit probe (save for the one which originally transmitted) must receive the signals. This procedure is represented in FIG. 2 as a two-step process. In this manner the sub-array will be filled and collection may proceed as outlined in paragraph 20. If this arrangement is made then the total number of firings needed and total number of waveforms needed are expressed in the equations below as
Number of firings(pitch-catch)=2(k+m1)1, and
Number of waveforms collected(pitch-catch)=k(2m1)+(m1).sup.2.

(10) In certain embodiments scan increment is equal to element pitch. In these embodiments an alternative to moving the probe along the direction of the array is to make a probe with many elements and generate a transmit and receive sequence which is equivalent to moving a smaller probe. While this requires the construction of a large array, it provides the advantage of potentially eliminating moving parts and positioning errors.

(11) Embodiments of the present invention may also be applied in the context of two-dimensional arrays, for which gains in data storage and firing efficiency can be even more dramatic. In fact, without application of the present, novel method to improve efficiency it is likely that matrix firing would be impractical to implement for all but the smallest two-dimensional arrays using computer technology available today. For example, consider the situation of a 16-element8-element array probe operating in pulse echo mode. Without implementation of such a technique to improve efficiency, (168)=128 firings would be needed at each probe position in order to obtain the (128129)/2=8256 waveforms. Assuming 1000 one-byte points per waveform and an array of 100 by 100 probe positions, this leads to a total data storage size of 82 GB.

(12) By re-using data collected at different probe positions the present invention provides a very considerable savings in number of firings and data collection. Consider a raster scan along both array directions for a two-dimensional array with n elements along the scan direction and m elements along the index (or step) direction. The raster scan includes k collection locations along the scan direction and 1 steps. An illustration of this arrangement is provided in FIG. 3. For the (nm)-element ultrasonic array operating in pulse-echo mode only m or n firings (whichever is smaller) are needed and (n.Math.m)+(n1).Math.(m1) waveforms need to be stored at each collection location. This corresponds to firing the corner element (i.e., element (1,1)) and receiving on all elements, then firing each element along the short edge of the array in turn (e.g., elements (M,1) for M=2 through m) and, for each firing, receiving on all elements along the long edge of the array (e.g., elements (1,N) where N=2 through n). (Additional waveforms will need to be recorded in order to fill data matrices near edges of the inspection grid). In one embodiment the probe will be scanned beyond the nominal scan boundaries and data will be taken such that data matrices near the boundaries are filled. The distance the probe is scanned beyond the scan boundaries is equal to the number of elements along each respective dimension minus one. Thus, if there are k scan points along the n dimension of the probe and l scan points along the m dimension of the probe, a conservative estimate of the total number of waveforms needed is shown below:
Number of waveforms(2D array)(k+n1).Math.(l+m1).Math.[n.Math.m+(n1).Math.(m1)].
This approximate formula is an overestimate of the total number of waveforms needed because fewer than [(n.Math.m)+(n1).Math.(m1)] waveforms will need to be collected at scan locations near the edges of the grid in order to provide a full reconstruction over the k by l grid. However, it is sufficient to demonstrate the advantage of use of this novel collection method in order to reduce data storage requirements. Returning to the 168 array example, this means that only 8 firings are necessary at each probe position and 233 waveforms need to be stored. For the same 100 by 100 array of probe positions this leads to total data storage of approximately (115.Math.107.Math.233).Math.1000 bytes equals approximately 3 GB which is easily achievable using the technology currently known in the art.

(13) In another embodiment of the present invention, as it pertains to two-dimensional arrays, symmetry is exploited only along the index direction. This may be done for a variety of reasons. As examples, if positioning along the index direction cannot be performed with sufficient precision, or if the index increment cannot be set equal to element pitch along that direction, or if the scan is performed along only one direction, then symmetry cannot be exploited along the index direction. In this case significant gains can still be made by implementing the following procedure: at each scan location, all m elements for which N=1 (i.e. element (M,1) where M=1 through m) are fired in turn, and for each firing all received waveforms from all elements in the array are recorded. (Note that, when firing element(s) (M,1) where M>1, waveforms at elements (M,1) where M<M does not need to be recorded because reciprocity considerations render it redundant.) At each scan location m firings are thus required and

(14) $\left[\left(m^2.Math.n\right)-\frac{m\left(m-1\right)}{2}\right]$
waveforms must be recorded. Extra waveforms will need to be recorded at locations beyond the nominal scan grid in order to record all waveforms needed to fill all matrices at every scan location. An estimate of the total number of waveforms needed for a scan over k by l locations is:

(15) $\mathrm{Number}\mathrm{of}\mathrm{waveforms}\left(2D\mathrm{array},\mathrm{symmetry}\mathrm{exploited}\mathrm{only}\mathrm{along}\mathrm{scan}\mathrm{direction}\right)\left(k+n-1\right).Math.l.Math.\left[\left(m^2.Math.n\right)-\frac{m.Math.\left(m-1\right)}{2}\right].$
This estimate is slightly conservative because it does not account for the reduced number of waveforms which need to be collected at locations beyond the nominal range k. Returning to the example of a 168 array with 100100 scan locations, the total data which needs to be stored is approximately (115.Math.100.Math.996).Math.1000 bytes equals 11.45 GB. This is still a very considerable improvement over the storage requirement of 82 GB which is required using the standard collection methodology.

(16) The present invention provides at least three advantages. The first is reduction of data storage. This allows more files to be stored on a single drive and also allows the ability in some cases to put all scan data into system memory, which would allow instantaneous access to all scan data. The second is potential for dramatically increased scan speed. Since less data is being acquired at each position, data throughput is reduced considerably. This increase in scan speed can result in reduced inspection costs. The third advantage is the potential for cleaner data because fewer transmitter firings results in a longer time interval between firings, which means that the sound has more time to dissipate.

(17) Alternatives to the present efficient data collection method involve collecting the full set of data at every scan location. This method results in slower scan times, potentially noisier data, and greatly increased (and in some cases impractical) storage requirements.

(18) While a specific embodiment of the invention has been shown and described in detail to illustrate the application of the principles of the invention, it will be understood that the invention may be embodied otherwise without departing from such principles.