SIMULATION OF B-SCAN IMAGES FROM TWO-DIMENSIONAL TRANSDUCER ARRAYS: PART II - COMPARISONS BETWEEN LINEAR AND TWO-DIMENSIONALPHASED ARRAYS

Transcription

1 ULTRASONIC IMAGING 14, (1992) SIMULATION OF B-SCAN IMAGES FROM TWO-DIMENSIONAL TRANSDUCER ARRAYS: PART II - COMPARISONS BETWEEN LINEAR AND TWO-DIMENSIONALPHASED ARRAYS Daniel H. Turnbull and F. Stuart Foste? Reichmann Research Building, Sunnybrook Health Science Centre Department of Medical Biophysics, University of Toronto 2075 Bayview Avenue, Toronto, Ontario, CANADA M4N 3M5 Two-dimensional (2-D) arrays have been proposed as a solution to the degradation in medical ultrasound image quabty occurring as a result of asymmetric focusing properties of linear phased array transducers. The 2-D phased transducer array is also capable of electronically steering the symmetrically focused ultrasound beam throughout a three-dimensional volume. In a companion paper the potential of 2-D transducer arrays for medical imaging has been investigated using simulated B-scan images. In this paper, the advantages of 2-D over linear transducer arrays is demonstrated by simulating images of spherical cysts embedded in a large scattering volume. The large elevation beamwidth in the neat-field of a 5 MHz linear phased transducer array results in a severe reduction in the image contrast measured between a 4 mm diameter cyst and the surrounding scattering media. By ernploying a 2-D array with symmetric focusing, the contrast between the cyst and surrounding scatterers is significantly improved. The use of additional elements in the elevation direction of a linear array is also investigated. In this case the additional elements are included only to focus, but not to steer the uhrasound beam. Using the contrast characteristics of a 4 mm diameter cyst, it is shown that relatively few elevation elements are required to significantly improve the nearfield imaging capability of the linear array Academic Press, Inc. Key words: B-scan image; contrast; elevation focusing; linear phased array; simulation; two-dimensional transducer array; ultrasound. I. INTRODUCTION Currently, linear phased array transducers are in wide use for medical ultrasound imaging. The linear phased array is limited to focusing and steering only in the azimuthal (array) direction. As a result of the asymmetrical focusing 1 Current Address: Toronto-Bayview Regional Cancer Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5. 2 Terry Fox Research Scientist of the National Cancer Institute of Canada /92 $5.00 Copyright by Academic Press. Inc. All rights of reproduction in any form reserved. 344

2 IMAGES FROM LINEAR AND 2-D ARRAYS properties of the linear array, the beam width in the elevation direction (perpendicular to array) can be much larger than the corresponding azimuthal beam width [l]. The symmetric focusing capability of a two-dimensional (2-D) phased transducer array, and the potential of the 2-D array to steer the focused beam throughout a three-dimensional volume has led to an interest in the use of such transducers for medical imaging [2-4]. In addition, recent studies indicate that 2-D transducer arrays may be necessary to implement correction schemes for phase aberration artifacts caused by tissue inhomogeneities [5,6]. In a companion paper [7], the imaging potential of two-dimensional (2-D) phased array transducers was investigated by simulating B-scan images of spherical lesions embedded in a large volume of randomly distributed point scatterers. In this paper, the same image simulation techniques are used to compare the imaging characteristics of a 2-D transducer array to those of a linear array having the same aperture and spacing between elements. Quantitative comparisons are made by measuring the contrast between a 4 mm diameter cyst embedded in a large random scattering volume. The effect of adding extra focusing elements in the elevation direction of a linear array is investigated by computing the beam properties and simulating images from various two-dimensional array geometries. II. COMPARISONS OF LINEAR AND 2-D TRANSDUCER ARRAYS A 101 element linear phased array transducer was simulated by using the 2-D array program described in a companion paper [7]. Starting with a 15 mm x 15 mm 2-D array with square elements of size 75 pm x 75 pm and 150 pm spacing between elements, the elements were tied together in the elevation direction using a constant 4 b) t Y (elevation) 101 variable delays k 4 15 mm... Fixed delay pattern on 101 elements simulating 60 mm cylindrical lens \ \ \ 60 mm \ \ -L Fig. 1 a) Geometry of 101 element linear array simulated via 2-D array program. Note that the fixed focus that would normally be used in the linear array is simulated by the feed delay pattern shown in b). 345

3 TURNBULL AND FOSTER pattern of time delays to provide a fixed cylindrical focus at 60 mm along the axis of the transducer as shown in figure 1. AppIication of variable delays in the azimuthal direction enables the simulation of a conventional linear array. In this section, comparisons are made between this 101 element Iinear array and the 2-D array having the same 15 mm x 15 mm square aperture. Impulse response- functions for the 101 element linear array, and the 101 x 1012-D transducer array were computed as described previously, using a 5 MHz ultrasound pulse [r]. In most medical imaging systems, the ultrasound beam is focused at one depth on transmit, and then dynamically focused on receive over a range of depths. Figure 2 depicts Y 4 t z Receive Beam b) -4Omm mm - c) - 40 mm----t - 60mm - Receive Beam Fig. 2 Elevation ultrasound beams of the 2-D and linear array transducers: a) 4 mm spherical cyst 60 mm from the transducers, where the transmit and receive beams are both focused; b) 2-D array imaging 4 mm cyst at 40 mm: transmit beam focused at 60 mm, receive beam focused at 40 mm; c) linear array imaging cyst at 40 mm: both transmit and receive beams focused at 60 mm. 346

4 IMAGES FROM LINEAR AND 2-D ARRAYS Axial Direction - a) W Contrast: f Fig. 3 Simulated 10 mm x 10 mm images of 4 mm spherical cyst: a) 101 x 1012-D array; b) 101 element linear array. Both transmit and receive focus at 60 mm. Contrast WAS measured in the 4 mm cyst as described in [7]: C = (S, - SJ / S, where Si = signal level measured in a circular region inside cyst and S, = signal level measured in same size region outside the cyst. schematically the elevation ultrasound beams of the linear and 2-D arrays in two situations: a) the transmit and receive beams are both focused at 60 mm (Fig. 2a); and b) the transmit beam is focused at 60 mm, while the receive beam is focused in the nearfield at 40 mm (Figs. 2b and 2~). The pulse-echo ultrasound beam is the convolution of the transmit and receive beams. At the elevation focus of the linear array (60 mm), the transmit and receive beams are both focused, and the beam properties of the linear array are similar to those of a 2-D array (Fig. 2a). This is confirmed in figure 3, where simulated images of a 4 mm spherical cyst placed 60 mm along the axis of the transducers (no steering) show the 2-D array and the linear array to be performing equivalently. When the cyst is moved to 40 mm, the 2-D array is still able to focus the receive beam in the region of the cyst (Fig. 2b), although the transmit beam is not focused so a deterioration in cyst contrast is expected. For the cyst at 40 mm, the linear array is unable to focus either the transmit or the receive beams in the elevation direction (Fig. 2~). Figure 4 shows simulated images of the 4 mm cyst placed at 40 mm, and the deterioration in cyst contrast from the linear array is marked. Figure 5 shows beam profiles of the 2-D and linear array transducers in the case of the transmit focus at 60 mm and the receive focus at 40 mm. The 2-D array has beam properties which are symmetric in the azimuthal and elevation directions (Fig. 5a), whereas the linear array has very different beam properties in the two directions (Fig. 5b). The beam width in the elevation direction is 3.5 mm compared

5 TURNBULL AND FOSTER Axial Direction - _,;-,.l_.l.._ 4 b) Contrast: 0.75 t ! Fig. 4 Simulated images of 4 mm spherical cyst for same arrays as in figure 3. Transmit focus at 60 mm; receive focus at 40 mm. to 1.2 mm in the azimuthal direction., and the peak response vaiues in this neat-field region are actually obtained off the axis of the transducer. Fiie 6 gives another demonstration that the large elevation beam width is -responsible for the inability of the linear array to detect the 4 mm cyst. In this case~the spberlcal cyst is rep&i&d by a 4 mm diameter cylindrical cyst having its axis in the elevzuion direction, which masks the large elevation beamwidth of the linear array tranadueer and dramatically improves the contrast between the cyst and the surrouuding scattering medium. III. ELEVATION FOCUSING The situation described above points out some of the inadequacies of linear array focusing. The problems of unacceptably large elevation beamwidth can be solved by providing extra elements for focusing in the elevation direction. Figure 7 shows an elevation-segmented linear array with a number of outlying elements in the elevation (y) direction. In 1982, Hassler et al. (81 developed such an array with a single elevation element on eitber side of the linear array (i.e., 3 elements per row) and demonstrated some improvements in foeusing. More recently, a group at General Electric has reported the fabrication of elevation-segmented linear arrays using so-called 2-2 PZT-polymer piezoeomposite materials, but they have not demonstrated the focusing properties of their array 191. In the elevation-segmented array, the number of elevation elements should be kept as small as possible. The purpose of the extra elements is not to steer the ultrasound beam, but -only to achieve improved foeusing properties. Therefore, such a transducer array should properly be thought of as 1.5-D : i.e., intermediate.between a linear and a 2-D transducer array. 348

6 IMAGES FROM LINEAR AND 2-D ARRAYS 10 a) 0 8 s E g-20 Lf 2-30 $ &I -40 Y z r b) O- S s Q-10 2 g-20 3 a g-30 - $ b-40 Y - x direction y direction z-5o -60' 'I L " I' I / I Thetao(deg) Fig. 5 Pulse-echo response vs 8 in azimuthal and elevation directions for transmit focus at 60 mm, receive focus at 40 mm: a) 101 x 1012-D array; b) 101 element linear array. Figure 8 shows the effect of subdividing the 101 element linear array into successively higher numbers of elevation elements. The elevation-segment boundaries ori, i= 1,2,...,N) were chosen as in a Fresnel zone plate, similar to Smith et al. [9]: Element i: Yi-I < IYI s Yi where yi = t/(i/n) A (1) 349

7 TURNBULL AND FOSTER Contrast: Fig. 6 Simulated image of 4 mm cylindrical cyst from 101 element linear Way. Transmit focus at 60 mm, receive focus at 40 mm (compare with figure 5b). Here A is half the elevation aperture, or 7.5 mm for this 15 mm x 15 mm array. The number of elevation time delays applied in the elevation-segmented array is N, the number of independent elements in the elevation direction (i.e., I central element and N-l pairs of outlying elevation elements). Thus, the total mm&r of independent elements or channels in the elevation-segmented array is N x 101. Figure 8a shows the beam profiles of the pulse-echo array response vs. 0 in the elevation direction for the array focused on transmit at 60 mm, and on receive at 40 mm. The elevation beamwidth decreases from 3.5 mm in the single element linear array to 1.2 mm in the elevation-segmented array with 4 segments. The beam profile (Lwatian) t (azitlhal) t4 4 * 101 elements (delays) Fig. 7 Elevation-segmented array. array designed to improve the elevation focusing of a linear 350

8 IMAGES FROM LINEAR AND 2-D ARRAYS a) Linear Array -. 2 Y-segments b) Linear 2-Y 3-Y 4-Y 2-D Contrast: 0.52tO $-O&I 0.69 f ?0.03 Fig. 8 a) Elevation beam profiles computed from a 15 mm x 15 mm, 101 element linear array, three elevation-segmented arrays, and the 101 x 101 element 2-D array. The transmit focus was at 60 mm, receive focus at 40 mm (on-axis). b) Simulated 10 mm x 10 mm B-scan images of a 4 mm spherical cyst from the same arrays as in a). of the 15 mm x 15 mm 2-D array with 101 x 101 elements is included for comparison. Figure 8b shows 10 mm x 10 mm simulated images of a 4 mm diameter spherical cyst, 40 mm from the arrays. The contrast between the cyst and the surrounding random scattering medium increases from 0.52 f 0.05 in the linear array to 0.69 f 0.04 in the array with 4 elevation-segments, which compares to 0.75 f 0.03, measured in the 2-D array image. 351

9 TURNBULL AND FOSTER These results indicate that using a small number of elevation elements in a :levation-segmented array, the nearfield focusing properties of linear phased arrays :an be significantly improved. The development of elevation-segmented arrays will necessarily involve increasing the total number of elements (channels) from the xurent limit to several hundred, after which technological advances may rave reached the point where a 1000 or more element 2-D array may be a realistic Ioal. Adding to the incentive to develop transducers of this type are reports showing bat two-dimensional transducers are necessary in order to implement correction xhemes to improve the deterioration of medical ultrasound image quality due the xesence of phase aberrations caused by tissue inhomogeneities [5,6]. [V. CONCLUSIONS The advantage of 2-D over linear phased transducer arrays has been demonstrated by showing the reduction of artifactual echoes inside the image of a 4 mm diameter cyst, caused by the large elevation beamwidth of the linear array. By iegmenting a linear array into a small number of elements in the elevation (y) direction, the neat-field focusing and imaging capability of the transducer improved. Simulated images have demonstrated a contrast improvement.n a nearfield image of a 4 mm diameter spherical cyst from 0.52 & 0.05 in a 101 :lement linear array, to 0.69 f 0.04 in an array with four elevation-segments, which zompares to 0.75 f 0.03 in the full 101 x 101 element 2-D array. These results.ndicate that by increasing the numbers of elements by a factor of 4 or 5, the focusing Troperties of an elevation-segmented array are equ&ient to those of a- full 2-D array with half wavelength spacing between elements. of course, the elevation-segmented transducer array can only steer the ultrasound beam in the a&muthai direction. If lrolumetric steering is required, then a fully two-dimensional transducer array must 3e developed. ACKNOWLEDGMENTS We are grateful to the National Cancer Institute of Canada for funding this research. D.H. Turnbull was supported by a University of Toronto Open Fellowship. Finally, we thank Michael Greenstein of Hewlett Packard Laboratories for useful discussions about this work. REFERENCES PI von Ramm, O.T., and Smith, SW., Beam steering with linear arrays, IEEE Trans. Biomedical Engineering BME-30,43&452 (1983). PI Turnbull, D.H., and Foster, F.S., Beam steering with pulsed two-dimensional transducer arrays, IEEE Trans. Ultrasonics Fewoelecttks Freq. Cwttrol38, (1991). [31 Smith, S.W., Pavy, H.G., and von Ramm, O.T., High-speed ultrasound volumetric imaging system - Part I: Transducer design and beam steering, IEEE Trans. Ultrada Fewoekcti Freq. Con& 38, (1991). 352

10 IMAGES FROM LINEAR AND 2-D ARRAYS 141 von Ramm, O.T., Smith, S.W., and Pavy, H.G., High-speed ultrasound volumetric imaging system - Part II: Parallel processing and image display, IEEE Trans. Ultraronics Ferroelectia Freq. Control 38, (1991). PI Trahey, G.E., and Freiburger, P.D., An evaluation of transducer design and algorithm performance for two-dimensional phase aberration correction, in 1991 IEEE U2fru.ronic.r Sjvnposium Proceedings, pp , (IEEE Cat. No. 91CH3079-1). PI O Donnell, M., and Li, P.C., Aberration correction on a two-dimensional anisotropic phased array, in 1991 IEEE ultrasonics Symposium Proceedings, pp , (IEEE Cat. No. 91CH3079-1). Tumbull, D.H., Lum, PK, Kerr, AT., and Foster, F.S., Simulation of B-scan images from two-dimensional transducer arrays: Part I - Methods and quantitative contrast measurements, Ukzronk Imaging, this issue. PI Hassler, D., Honig, D., and Schwarz, R., Uhrasound B-scanner with multi-line array, Ultrasonic Imaging 4,32-43 (1982). [91 Smith, L.S., Engeler, W.E., O Donnell, M., and Piel, J.L., Rectilinear phased array transducer using 2-2 ceramic-polymer composite, in 1990 IEEE Ultrasonics Symposium Proceedings, pp , (IEEE Cat. No. 9OCH2938-9). 353