- Email: [email protected]
A 20-MHZ ultrasound system for imaging the intestinal wall
Ultrasound in Med. & Biol. Vol. 15, No. 3, pp. 273-280, 1989
0301-5629/89 $3.00 + .00 © 1989 Pergamon Press plc
Printed in the U.S.A,
OOriginal Contribution I
A 20-MHZ ULTRASOUND SYSTEM FOR IMAGING T H E I N T E S T I N A L WALLer ROY W. MARTIN, FRED E. SILVERSTEIN* a n d MICHAEL B. KIMMEY* Department of Anesthesiology and Center for Bioengineering and ,Department of Internal Medicine, Division of Gastroenterology, University of Washington, Seattle, WA 98195
(Received21 April 1988; in final form 3 August 1988) A b s t r a c t - - A n ultrasound system has been developed which uses high-frequency (20 MHz) ultrasound to provide
high-resolution images of tissue. The system provides 0.21-ram range and 0.65-ram lateral resolution. The transducer aperture size is 1.8 mm maximum. Miniature probes have been developed which can image via the biopsy channels of standard fiberoptic endoscopes as well as probes for imaging in vitro. A commercially available video "frame grabber" is used in conjunction with a standard microcomputer for image acquisition. This allows images to be displayed and recorded on standard television equipment and be stored and manipulated digitally. The features of the system allow in vivo ima~ag, in vitro intagi'ng after resection, and histological images of the same tissue region to be acquired and compared. This method is particularly useful in learning how to correctly interpret ultrasonic images of the intestinal wall. The use of 20 MHz is advantageous in achieving excellent resolution and small size probes. The system provides a unique approach to imaging the intestinal wall. Key Words: Ultrasound, Intestinal wall Imaging, Miniature probes, High resolution, Endoscope probes, Digital ultrasound, Histological comparison.
INTRODUCTION
emerged (Tytgat and Tio 1986). This advantage is the potential of the method in diagnosing diseases of the wall of the intestinal tract, an area where whole body imaging devices have inadequate resolution and localization. The endoscope can be used to locate a lesion, image principally the surface and then the ultrasound can be applied to examine the deeper structure of the lesion in cross section. As investigators began focusing on the acoustical appearance of the intestinal wall it became clear to us that careful in vitro studies were needed of intestinal wall tissue in order to learn how to interpret the images properly (Silverstein et al. 1986; Kimmey et al. 1987, 1988). Because the wall itself was the imaging target the depth of penetration needed was much less than that required to image adjacent organs. We therefore hypothesized that frequencies higher than l0 MHz could be used which would provide better resolution to delineate the layers of the wall. Higher frequency would also allow use of small aperture probes. We therefore developed a 20-MHz imaging system which could be used both in vitro, for imaging excised tissue and making histological comparisons, and in vivo, with a new endoscopic echo probe (Silverstein et al. 1989). This new endoscopic probe was designed so it did not require the ultrasonic trans-
Combined ultrasound and fiberoptic endoscopes were initially developed a decade ago in an attempt to provide improved ultrasonic visualization of organs adjacent to the intestinal tract such as the pancreas (DiMagno et al. 1982). Locating the ultrasound transducer on the tip of a gastroscope allowed close proximity imaging of these organs, offering the possibility of using high frequency (7.5-10 MHz) to provide improved image resolution. Developments in combined ultrasonic and optical viewing endoscopes continue and include the application for the diagnosis of thoracic and abdominal disease (Ohmori et al. 1983; Sivak 1987; DiMagno et al. 1982; Tio and Tytgat 1986; Yasuda et al. 1988). However, developments in cross-sectional imaging methods (e.g., MR and CT) have been dramatic and high-quality images of abdominal and thoracic organs are now available. Notwithstanding these developments, a unique advantage of combining ultrasound and endoscopy has t This work was supported by NIH grant #DK 34814. Address all correspondence to: Roy W. Martin, Ph.D., Department of Anesthesiology and Center for Bioengineering. RN-10, University of Washington, Seattle, WA 98195. 273
274
Ultrasound in Medicineand Biology
Figure 1 shows the method as applied to an apparatus for in vitro scanning of excised tissue. Here, the probe is held in a fixture with the transducer located below the surface of the solution in close proximity to the tissue. As it scans horizontally along the mounted tissue, the echo ranges in a vertical direction through the tissue. A potentiometer attached to the scanning apparatus transduces the horizontal movement of the probe and controls the location of each scan line to properly represent the spatial position. The scanning action generates a rectilinear image compatible with standard television format, allowing the recording and display of images with standard TV recorders and monitors using the modified video frame acquisition circuitry (discussed below). However, the vertical echo ranging is assigned to the horizontal axis used in TV since the time required to produce a horizontal TV line (63.5 #s) is more compatible with time required to echo range through the tissue (26 #s for a 20-mm depth). For that reason, we assigned the horizontal probe scanning movement to the vertical axis of the TV. A block diagram of the system is shown in Fig. 2. It is built around an IBM computer (IBM-AT) and a frame grabber board (Oculus 200, Coreco Inc., Longueuil, Quebec, Canada) which fits into the computer. The frame grabber is a key component of the system, since it provides the video frame acquisition (i.e., it provides a means of capturing an image and displaying it on standard video equipment). The frame grabber contains a 7 bit analog to digital converter with fast access memory, accepting video information at a rate of l0 MHz and storing a complete
position adjustment knob (rotationproduces x directiontranslation) y
X
-
~
,.i..~t. modifiS d micro manipulation
y tissue
dish
Volume15, Number 3, 1989
positionsensing transducer tissue mountingblock
Fig. 1. A miniature ultrasonic probe is mounted on a modified micromanipulator for scanning a tissue specimen located in a dish of saline. A U'ansducer is added to detect the X axis movement which becomes the vertical axis on a TV
monitor. The ultrasonic probe sends energy in the Y direction and obtains echoes displayed on the horizontal axis of the TV monitor. ducer to be a part of the fiberoptic endoscope but rather an accessory to it, an important practical advantage. We report the engineering aspects of the system, provide specifications of it and give some examples of images obtained with the new system. More thorough evaluation of its use in vivo is being reported elsewhere. SYSTEM DESCRIPTION The imaging concept of the unit is simple since it functions with linear mechanically scanned probes.
Ultrasonic transceiver
Panasonic
video monitor
Ultrasonic~" ~ - ~
l[ displacement I Displacement transducer
i
Coreco Oculus 200 video frame grabber
Probe I
]E A
v
1
Timing and interface
I
Fig 2. Block diagram of the electronic section of the 20-MHz system which operates with either an in vitro echo probe or an in vivo endoscopic echo probe. The image formation and display output occurs in the video frame grabber. It receives the echo information from the transceiver, timing and control from the timing and interface section, and the computer.
A 20-MHz ultrasound systemfor imagingthe intestinalwall • R. W. MARTINel aL
512 × 480 pixel (7 bits per pixel) image. Further, the computer can operate on this image to modify, magnify, or enhance it via special image processing and manipulation programs provided with the frame grabber (Gray Library, Coreco Inc.) and via some special programs we have developed. This program library consists of approximately 75 subroutines which can be called from the C language. The image captured by the frame grabber is continuously displayed on the black and white monitor (Fig. 2), and any manipulation to it performed by the computer immediately appears in the image. The frame grabber was designed to operate with video cameras, and we have retained this feature as a switch selectable function to use in recording video images of histology obtained with a video microscope. However, for ultrasonic imaging, we modified the frame grabber and linked it to the rest of our system so that ultrasonic echo and ultrasonic transducer position information could be used to generate an image. This refinement required minor modifications to the Coreco board, as well as building the timing and interface circuit indicated in Fig. 2. This timing and interface board (details in Fig. 3) provides the coordination between the frame grabber, the ultrasonic transceiver, and the probe displacement transducer. This displacement transducer senses the position of the probe and is used to insure that each scan line is properly positioned.
The key elements of the method of controlling the frame grabber to acquire, store, and display images are shown in Fig. 3. Minor modifications were made to the Coreco board to allow interrupting the normal control of writing to memory and the memory row addressing. We accomplished the first task by routing the write-enable signal (WR) in the Coreco board to the control and timing of our timing and interface board. We then resubmitted it to the Coreco board as the WRE signal but timed in the manner we required. Next, we replaced the Bus Memory Address Multiplexer chips in the Coreco board with chips similar in function but with outputs that can be disabled. The inputs and outputs of these replaced chips were connected in parallel to identical chips in our timing and interface board. The control and timing circuit in our board disables and enables (BEN1 and BEN2 signals) these Bus Multiplexers to switch the address control between the Coreco board and our timing and interface board. When acquiring an ultrasonic image the addressing of the memory, during the write to memory cycle, is switched to the timing and interface board. There the row addressing is taken from the output of the probe displacement transducer. This displacement signal is converted to a digital number by the analog to digital convener (Fig. 3) and then latched and held during the time each scan line is acquired. This digital number is in the correct format to directly address the memory and force the
CORECO OCULUS 200 BOARD Row Bus and multi )lexor column at ess s __ 512 x 480 x 8 Bit image "--1 ---~ memory
i Control and timing
l
*
WRE
,.u,re image
Probe displacement signal
= Control I and timin!
BEN 1
I
BEN 2 ;
Row address ~
multiplexor
Latch TIMING
275
AND INTERFACE BOARD
Fig. 3. Block diagram of the key elements of our methodology of controlling the frame grabber to acquire, store, and display images are shown. The abbreviations are: WR = write to memory command signal, WRE = write to memory enable, BEN 1 = bus # 1 enable, BEN2 = bus #2 enable, EL = enable latch, and A/D = analog to digital COrlverter.
276
Ultrasound in Medicine and Biology
Volume 15, Number 3, 1989
digitized echo signal to be written to the correct position to generate the proper B mode line. We developed a method to improve the sampling resolution established and limited by the sampling rate of the frame grabber. The restriction is the l0 MHz sampling rate of the analog to digital converter in the Coreco 200 frame grabber. We doubled this rate by using a technique similar to that employed in sampling oscilloscopes. Considering first the method for generating a scan line, the transducer is excited and the echoes resulting from this transmission are digitized and stored in the frame grabber as a partial B mode line. The digitization at the l0 MHz sampling rate is synchronized to the positive phase of the transducer excitation signal. Next, after the acquisition of the first set of echoes is complete, but before the completion of the horizontal sweep, the transmitter is again activated. This time, the beginning of the acquisition is delayed one half of a 10 MHz cycle. This second set of information is stored sequentially as another part of the B mode line. Because of this phase delay, sampling from this acquisition permits sampling between the former sampling points. The net effect is that information is acquired at twice the number of spatial points and two images are acquired this way. The final step is to merge the two B mode images together to form one image. This merger is accomplished by placing each sequentially acquired pixel in the second image after the corre-
[I
same.
Consequently, the transceiver consists of two identical transmitting and receiving channels whose output is multiplexed to a common detector circuit. The envelope of the RF echo data is extracted by this last circuit and provided as the output to the frame grabber. The timing, control and time varying gain control section is dual in nature, in that it provides the transmit enable commands (EN 1 and EN2), the
ULTRASONIC TRANSCEIVER
I Zone 1 I Matching transducer I M1 I
sponding pixel acquired in the first image. The result is an image line with twice the number of points and spatial sampling every 0.035 mm along the radiation path. The ultrasonic transceiver, the second key component of the system, operates with the ultrasonic transducer(s) to acquire echo information from the tissue (Fig. 2). A block diagram of the transceiver appears in Fig. 4. The system has two alternate modes, either single or dual channel. In single mode, one transducer is used to image the entire field of interest ranging 1-20 m m from the transducer. In dual mode, two transducers are used. One element is optimized for imaging a field adjacent to the transducer (1-5 mm); the second, for imaging a field extending from 5-20 mm. Images from both fields are acquired line by line by switching between channels, then combining them to produce a composite image. In the single channel mode the system is locked to one channel, but the rest of the system functions the
Transceiver #1
.....H
Channel T. .I.jG1}'~ multiplexor EN1
I
transducer
I
I Adjust ~ TVG CH1 & CH2 I base and slope gain ! ! I I !
single or dual L channel mode
Detector Ultrasonic echo signal I
M2
_~ ~ I I LI
|
I
.
EN21 I
Timing control and TTme varying gain_ control ('I'VG)
I
I- -
[TVG2 I ~ ~ I
[ I I EN1 - ~ TVG1
I
I I I I ! ! To/From Fr.a~negrabber I Timing and interface
Timing/Synch j
Fig. 4. A block diagram o f the ultrasonic transceiver is shown. The unit consists o f two identical transmitting and receiving channels whose outputs are multiplexed to a c o m m o n detector circuit. The envelope of the R F echo data is extracted by this last circuit and provided as the output to the frame grabber.
A 20-MHzultrasoundsystemfor imagingthe intestinalwall• R. W. MARTINel al. time varying gain control (TVGI and TVG2), and the channel multiplexer control (CHS). PROBE DESCRIPTION We have built eleven 20 MHz probes of various types: eight for in vitro studies and three for endoscopic use. All the probes use gold plated PZT5 piezoelectric material obtained from Valpey-Fisher Corporation, Hopkinton, MA. Miniature disk elements were cut from larger disks using miniature diamond core drills (Lindsay Enterprises Inc., Seattle, WA). The elements are operated in the thickness mode and are backed with air over most of their back face. The diameter of elements ranges between 1.0 and 1.8 ram. Electrical connection is by soldering (Martin et al. 1988), and the front face is coated with a bio-compatible epoxy (Epoxylite 6001, Epoxylite Co., South E1 Monte, CA) after the elements are connected and mounted. In vitro probes are mounted in an assembly that allows easy interchange with other ultrasonic transducers for comparison studies.
/3
The endoscopic imaging probes are designed to be used as an accessory to a fiberoptic endoscope. First, they are built small enough to be passed through the biopsy channel of the endoscope. Second, the probe is constructed so that the tip can be moved along the intestinal wall by sliding the probe in and out of the endoscope. The tip movement is detected by a displacement transducer that is readily attached to the hand piece of the endoscope. Thirdly, the ultrasonic transducer is mounted at the probe tip so it radiates perpendicular to the probe's axis. The signal from the displacement transducer and the echo signals from the ultrasonic transducer as it is moved along the wall allows the system to produce the desired B mode image. Details of the design of the tip are given in Fig. 5. The design allows the ultrasonic element in the tip to be rotated so that the transducer's radiation beam is normal to the intestinal wall. The outer body of the major length of the probe consists of a thin wall Teflon catheter of 2-m length. Inside this catheter is a special cable designed to have high flexibility but
(6w)
(.350)
2m
8.9 mm
/Teflon /tubing mm OD
Glued on I silk I / thread,
/Painted & /coated / w i t h Epoxylite
~
//Needle
ooo 8888o888)888'888
xiglass
~
o
~'/.,"A~/.AI'e'¢,,¢,¢,¢~.~I ~
Triple spring 1.00 mm ID
~l
~
AT 0 0 0 0 0 0 O_0 " ~ ~ 0 0 0 'O00~\".\V'~k\X
\ 'Stainless. \ .~\ \ X \ steelstrain Crimped 50 ohm \a,~,,.~¢~,, re,e) tubing ~wu.~.,,. v..eT___ coax coup,,.ng .46 rnm OD ~lllcone
sheath
277
material
20 mHz'~ Plexiglass 'IgiSm)m xta xtal xtal mount ,.s 1.8 mm OD
3 mm OD
Fig. 5. A cross-sectional drawing of the tip of the endoscopic probe used for in vivo scanning is shown. The 20-MHz piezoelectric clement radiates down in this illustration. An acoustical coupling media is included between the surface of the transducer and the surface of the probe. This coupling provides a delay before the signal encounters the mucosa when the probe is placed directly on the mucosa. The delay prevents the echo signal from being obscured by the ring-down from the excitation pulse. A silicone sheath is used to seal the probe tip from the teflon tubing so that fluids from the bowel will not leak around the inner cable of the catheter. The sheath is flexibleenough to allow the tip to be rotated with respect to the teflon tube over a limited range (two full 360° turns). The inner cable attaches to the tip assembly as illustrated. When the cable is rotated, the tip assembly also rotates. This rotational action allows orientation of the transducer towards the intestinal wall to obtain the optimal cross-sectional image of the wall. The inner coaxial wire connects to the piezoelectric element as shown.
278
Ultrasound in Medicine and Biology
Volume 15, Number 3, 1989
good torsional properties (Martin et al. 1988). This cable is used to transmit rotational force from the proximal end to the piezoelectric element located on the distal tip. Located inside this cable are the electrical wires which connect to the transducer. SPECIFICATIONS OF THE SYSTEM Special computer operations of the system include a program to read in and merge the ultrasonic images, programs to store or recall images from either floppy or hard disk memory and a program to reduce four single full screen images into a composite of four in one screen image. A program also allows outlining regions in images to enable calculation of linear dimensions and areas which can then be compared quantitatively. The band width of the receiver is 5 MHz; the noise referred to the input is 8-tzV rms, and the maximum gain is 85 dB. The time varying gain provides 70 dB of gain control. The receivers optimal noise source impedance (Motchenbacher and Fitchen 1973) is 50 r, as is the transmitter's output impedance. The maximum transmitted voltage into a 50 fl load is 50 Vpp for a 5.5 cycle burst. The typical impedance at 20 MHz of a transducer element connected by 2-m cable of 50 f~ characteristic impedance is 80 + j26 ft. We electrically match the transducers to the system's 50 fl impedance to optimize performance. The electrical to acoustic transfer of one of the endoscopic ultrasonic probes (single element 1.8 mm in diameter) was measured in the following way. A 50 fl output impedance linear power amplifier (RF Communication Inc. #805) was driven by a gated generator (Wavetek Inc. # 164) so it would produce 5 cycles of a 20 MHz sine wave at a repetition rate of 100 kHz. The generator was adjusted so the power amplifier applied 38 Vpp into a 50-fl load. The resistive load was then disconnected and the probe was connected to the power amplifier. The probe was then submerged in water and was directed at the absorber of a Nelson Radiometer #SA-200-002 (NTR Systems, Inc., Seattle, WA). The radiated acoustic power was then measured and the applied electrical power computed. The ratio of electrical to acoustic transfer was then calculated giving a value of -18.6 dB. The 6-dB radiation pattern (Fig. 6) of one of the probes was obtained by imaging seven 0.15-mm diameter nylon strings in a water bath. The image was acquired by first setting the gain appropriately for the experiment. A 3-dB attenuator circuit was introduced
Fig. 6. Image made with the 20-MHz system by scanning 0.15-mm diameter nylon strings is shown. The transducer
was a single disk 1.8 mm in diameter. These strings are suspended in a fluid filled tank, 2 mm apart in both the horizontal and the vertical direction. The vertical spots at the upper fight of the photographs arose from vertically arranged strings. Other artifacts in the image are produced by reflections from the tank walls.
between the transceiver and the transducer at a point were both where impedance matched to each other and to the attenuator. This action produced a 6-dB attenuation of echoes since the transmitted energy passed twice through the network, first during excitation and second when the reflected energy passed into the receiver. The gain of the receiver was then adjusted while scanning until the strings were barely visible in the image. The attenuation was then removed and the strings were scanned and the image recorded. This recorded image represents all the echoes that had amplitudes within 6 dB of the peak point or central part of the beam and therefore represents the 6-dB beamwidth pattern. The image shown (Fig. 6) is from a single disk 1.8 mm in diameter; the narrowest part of its beam occurs at 9 mm from the probe where the lateral resolution is 0.45 mm. The largest lateral beamwidth in the shown imaging field is 0.85 mm. The thickness of the string target was subtracted in making this determination. For a 1-mm diameter element the best lateral resolution measured with the same string target was 0.65 m m and that occurred at a distance of 5 mm from the probe. The average probe range resolution is 0.21 mm.
A 20-MHz ultrasound system for imaging the intestinal wall • R. W. MARTIN el al.
279
IMAGING EXAMPLES Excised intestinal wall tissues have been imaged in vitro with the 20-MHz LEP probe. An example of an in vitro image of canine cecum acquired with our system is shown in Fig. 7. All the wall layers are clearly evident in this image. The histological image shown was taken from approximately the same region. There is good agreement in the layered appearance seen in the ultrasonic and histological images. Of particuiar note in this image is the thickness of the mucosal region. In vivo studies have been conducted in anesthesized dogs. An example of an image acquired of the esophageal wall at the level where the aorta runs adjacent to the esophagus is shown in Fig. 8. The aorta is visible in this image as well as the cross section of an additional vessel. Note that the mucosa appears to blend with the submucosa in Fig. 8. In man, the mucosa of the esophagus is known to be thinner than the mucosa of the stomach (0.3 mm in the esophagus and up to 1.5 m m in the stomach (Weiss and Greep 1977). The mucosa is too thin to appear as a separate layer with our current system.
rn sm
mp s
Fig. 8. Image of the normal canine esophageal wall acquired in vivo with the 20-MHz echo probe system. The vertical axis represents distance into the tissue with the transducer located in the esophagus in the upper part of the image (full scale = 1.8 cm). Horizontal axis is distance along the esophagus (full scale = 2 cm). Note the presence of two vessels adjacent to the wall. EA---excitation artifact, M--mucosa, SM--submucosa, MP--muscularis propria, and S--adventitia, IVW--inner or proximal wall of the blood vessel, OVW~outer or distal wall of the aorta, AO --aorta, and BV--blood vessel.
The dark layer below the bright mucosa and submucosal layer in the image is esophageal muscularis propia, the second bright layer includes the adventitia and the blood vessel wall. The total wall thickness in the esophagus of the dog has been reported to be approximately 4 mm in the cervical region, 2.5 mm in the thoracic region, and 6 mm in the abdominal region (Evans and Christensen 1979). Since the image was taken in the thoracic region the range of wall thickness of 2-3 mm appearing in the image of Fig. 8 agrees with this report. DISCUSSION
ffl
sit3
mp s
Fig. 7. Upper photograph. An ultrasound image taken in vitro of an excised section of canine cecum: vertical axis = depth into the tissue (full scale = 6 mm); horizontal axis = distance along the tissue (full scale = 1.8 cm). The layers seen on the image (from top to bottom) m--mucosa, sm-submucosa, mpumusculads propria, and s--serosa. Lower photograph. Histology of the same specimen taken in close proximity to the region ultrasonically imaged and scaled to approximately the same size. Note: The ultrasonic image was taken approximately 30° oblique to vertical axis in the photograph. It is oriented as shown for ease in comparison to the histological image.
The computer-basod imaging system we have developed is versatile for researching a variety of questions. It is readily applicable to methods of computer processing of ultrasonic images. Computerized comparisons of histology and ultrasonic images are possible if the histological images are acquired through a video camera since they then can be read into the computer via the frame grabber. The radiation pattern found for the 1.8-mm diameter element agrees quite closely with theory in three ways. First, using the usual equation (Wells 1977) for the near-field and far-field transition point (z' = a2f/c where a = radius of a disk transducer, f = frequency, and c = the velocity of propagation of sound) this transition occurs in water for a 1.8-mm diameter transducer at 11 mm. The narrowest part of a one way 3-dB beamwidth for disk transducers exci-
280
Ultrasound in Medicine and Biology
rated with more than 2 cycles (Weyns 1980) occurs closer to the transducer than z'. We found it occurred at 9 mm for the 1.8-mm diameter probe which agrees with Weyns' simulations. Second, the far-field one way 3-dB beamwidth divergence angle may be calculated from the directivity function (Kinsler et al. 1982) and is given by the arc sin(O.318c/fa). This angle then may be used to calculate a transmission and reflection 6-dB beamwidth pattern at various points. For a 1.8-mm element at a distance of 15 mm this gives a width of 0.8 mm which compares closely with the 0.85 mm we measured. Finally, for a transducer that has a wide enough bandwidth to pass an excitation tone burst with an envelope which approximates a window function, the range resolution will be equal to cT/2 where T = the tone burst duration. Our excitation is 5.5 cycles at 20 MHz which when applied to this equation gives 0.2-ram range resolution in water. This is close to what we measured from the string targets after subtracting the thickness of the string from the echoes. The l-ram diameter transducer also agreed fairly close to the theory except for the point at which the narrowest beam occurred. We measured a distance of 5 mm from the transducer whereas theory predicts it should be less than 3.5 mm. However, we cannot be certain of our measurements because the string pattern spacing of 2 mm may have been too coarse to show the pattern properly. The images that we have acquired with the system illustrate several advantages. First, images of normal intestinal wall can be obtained that correlate closely with anatomical layers. Second, the fact that images were acquired with a single element 1.8-mm diameter or less from a nonfocused aperture is important from the viewpoint of building ultrasonic imaging probes that can be passed through the biopsy channels of endoscopes. The high frequency of 20 MHz opens the door to small aperture transducer design although the depth of penetration may be a limiting factor if the wall is thickened. Several future improvements could be made to the system. First, it would be useful to interface the computer more closely to the control of image acquisition, thus providing ease in changing or modifying scanning sequences. Second, the addition of logarithmic compression in the receiver would be advantageous in improving the quality of the image since only time-varying gain compensation was used in the images shown. Finally, designing the system so it could operate over a wide range of ultrasonic frequencies may make it possible to study a wider range of intestinal wall pathologic conditions.
Volume 15, Number 3, 1989
In summary, the system we have developed allows imaging of the intestinal wall both in vitro and in vivo with high resolution. This dual ability is useful in conducting wall imaging research. Although improvements in the system for in vivo imaging are anticipated, it may be possible in its present form for diagnosis of many forms of intestinal wall disease. AcknowledgmentmWe thank and acknowledge Andy Proctor for his work in building probes, Yehuda Sabag in electronic fabrication, and Don Franklin in ultrasonic imaging and preparing the histological image.
REFERENCES Di Magno, E. P.; Regan, P. T.; Clain, J. E.; James, E. M.; Buston, J. L. Human endoscopic ultrasonography. Gastroenterology 83:824-829; 1982. Evans, H. E.; Christensen, G. C. Miller's anatomy of the dog. 2nd ed. Philadelphia: W. B. Saunders Co.; 1979:456. Kimmey, M. B.; Silverstein, F. E.; Haggitt, R. C.; Shuman, W. P.; Mack, L. A.; Rohrmann, C. A.; Moss, A. A.; Franklin, D. W. Cross sectional imaging method: A system to compare ultrasound (US), computed tomograph (CT), and magnetic resonance (MR) with histological findings. Invest. Radiol. 22:227231; 1987. Kimmey, M. B.; Silverstein, F. E.; Martin, R. W. Ultrasound interaction with the intestinal wall. In: Kawati, K., ed. Endoscopic ultrasonography in gastroenterology. New York: IgakuShoin; 1988:35-43. Kinsler, L. E.; Frey, A. R.; Coppens, A. B.; Sanders, J. V. Fundamentals of acoustics. New York: John Wiley & Sons; 1982:180. Martin, R. W.; Silverstein, F. E.; Kimmey, M.; Jiranek, G.; Proctor, A. B mode imaging and Doppler ultrasonic catheters for use with fiber optic endoscopes. Proceedings of the International Society of Optical Engineering: conference on microsensors and catheter based imaging technology. 1988; 904:121-126. Motchenbacher, C. D.; Fitchen, F. C. Low-noise electronic design. New York: John Wiley & Sons; 1973:35-37. Ohmori, S.; Giorgi, F. J.; Starosta, M. An ultrasound imaging system for invasive applications. Proceedings of IEEE: Frontiers of engineering and computing in health care. IEEE CH 1896; 1983:457--460. Silverstein, F. E.; Kimmey, M. B.; Martin, R. W.; Haggitt, R. C.; Moss, A. A.; Mack, L. A.; Franklin, D. W. Ultrasound and the intestinal wall. Scand. J. Gastroenterol. 21(Suppl. 123):34-40; 1986. Silverstein, F. E.; Martin, R. W.; Kimmey, M. B.; Jiranek, G. C.; Franklin, D. W.; Proctor, A. Experimental evaluation of an endoscopic ultrasound probe: in vitro and in vivo canine studies. Gastroenterology [ 1989]. Sivak, M. V. Special methods and techniques in gastroenterologic endoscopy. In: Sivak, M. V., ed. Gastroenterologic endoscopy. Philadelphia: W. B. Saunders Co.; 1987:181-202. Tip, T. L.; Tytgat, G. N. J. Atlas of transintestinal ultrasound. The Netherlands: Mur-Kostvefloren, Aalsmere; 1986. Tytgat, G. N. J.; Tip, T. L., editors. Proceedings of the 4th International Symposium on Endoscopic Ultrasonography. Scan. J. Gastroenterol. 21(Suppl. 123); 1986. Weiss, L.; Greep, R. O. Histology. 4th ed. New York: McGrawHill; 1977:665-668. Wells, P. N. T. Biomedical ultrasonics. London: Academic Press; 1977:28. Weyns, A. Radiation field calculations of pulsed ultrasonic transducers Part 1: planar circular, square and annular transducers. Ultrasonics 18(4):183-188; 1980. Yasuda, K.; Hidekazu, M.; Fujimoto, S.; Nakajima, M.; Kawai, K. The diagnosis of pancreatic cancer by endoscopic ultraoniography. Gastrointest. Endosc. 34(1):1-8; 1988.
0301-5629/89 $3.00 + .00 © 1989 Pergamon Press plc
Printed in the U.S.A,
OOriginal Contribution I
A 20-MHZ ULTRASOUND SYSTEM FOR IMAGING T H E I N T E S T I N A L WALLer ROY W. MARTIN, FRED E. SILVERSTEIN* a n d MICHAEL B. KIMMEY* Department of Anesthesiology and Center for Bioengineering and ,Department of Internal Medicine, Division of Gastroenterology, University of Washington, Seattle, WA 98195
(Received21 April 1988; in final form 3 August 1988) A b s t r a c t - - A n ultrasound system has been developed which uses high-frequency (20 MHz) ultrasound to provide
high-resolution images of tissue. The system provides 0.21-ram range and 0.65-ram lateral resolution. The transducer aperture size is 1.8 mm maximum. Miniature probes have been developed which can image via the biopsy channels of standard fiberoptic endoscopes as well as probes for imaging in vitro. A commercially available video "frame grabber" is used in conjunction with a standard microcomputer for image acquisition. This allows images to be displayed and recorded on standard television equipment and be stored and manipulated digitally. The features of the system allow in vivo ima~ag, in vitro intagi'ng after resection, and histological images of the same tissue region to be acquired and compared. This method is particularly useful in learning how to correctly interpret ultrasonic images of the intestinal wall. The use of 20 MHz is advantageous in achieving excellent resolution and small size probes. The system provides a unique approach to imaging the intestinal wall. Key Words: Ultrasound, Intestinal wall Imaging, Miniature probes, High resolution, Endoscope probes, Digital ultrasound, Histological comparison.
INTRODUCTION
emerged (Tytgat and Tio 1986). This advantage is the potential of the method in diagnosing diseases of the wall of the intestinal tract, an area where whole body imaging devices have inadequate resolution and localization. The endoscope can be used to locate a lesion, image principally the surface and then the ultrasound can be applied to examine the deeper structure of the lesion in cross section. As investigators began focusing on the acoustical appearance of the intestinal wall it became clear to us that careful in vitro studies were needed of intestinal wall tissue in order to learn how to interpret the images properly (Silverstein et al. 1986; Kimmey et al. 1987, 1988). Because the wall itself was the imaging target the depth of penetration needed was much less than that required to image adjacent organs. We therefore hypothesized that frequencies higher than l0 MHz could be used which would provide better resolution to delineate the layers of the wall. Higher frequency would also allow use of small aperture probes. We therefore developed a 20-MHz imaging system which could be used both in vitro, for imaging excised tissue and making histological comparisons, and in vivo, with a new endoscopic echo probe (Silverstein et al. 1989). This new endoscopic probe was designed so it did not require the ultrasonic trans-
Combined ultrasound and fiberoptic endoscopes were initially developed a decade ago in an attempt to provide improved ultrasonic visualization of organs adjacent to the intestinal tract such as the pancreas (DiMagno et al. 1982). Locating the ultrasound transducer on the tip of a gastroscope allowed close proximity imaging of these organs, offering the possibility of using high frequency (7.5-10 MHz) to provide improved image resolution. Developments in combined ultrasonic and optical viewing endoscopes continue and include the application for the diagnosis of thoracic and abdominal disease (Ohmori et al. 1983; Sivak 1987; DiMagno et al. 1982; Tio and Tytgat 1986; Yasuda et al. 1988). However, developments in cross-sectional imaging methods (e.g., MR and CT) have been dramatic and high-quality images of abdominal and thoracic organs are now available. Notwithstanding these developments, a unique advantage of combining ultrasound and endoscopy has t This work was supported by NIH grant #DK 34814. Address all correspondence to: Roy W. Martin, Ph.D., Department of Anesthesiology and Center for Bioengineering. RN-10, University of Washington, Seattle, WA 98195. 273
274
Ultrasound in Medicineand Biology
Figure 1 shows the method as applied to an apparatus for in vitro scanning of excised tissue. Here, the probe is held in a fixture with the transducer located below the surface of the solution in close proximity to the tissue. As it scans horizontally along the mounted tissue, the echo ranges in a vertical direction through the tissue. A potentiometer attached to the scanning apparatus transduces the horizontal movement of the probe and controls the location of each scan line to properly represent the spatial position. The scanning action generates a rectilinear image compatible with standard television format, allowing the recording and display of images with standard TV recorders and monitors using the modified video frame acquisition circuitry (discussed below). However, the vertical echo ranging is assigned to the horizontal axis used in TV since the time required to produce a horizontal TV line (63.5 #s) is more compatible with time required to echo range through the tissue (26 #s for a 20-mm depth). For that reason, we assigned the horizontal probe scanning movement to the vertical axis of the TV. A block diagram of the system is shown in Fig. 2. It is built around an IBM computer (IBM-AT) and a frame grabber board (Oculus 200, Coreco Inc., Longueuil, Quebec, Canada) which fits into the computer. The frame grabber is a key component of the system, since it provides the video frame acquisition (i.e., it provides a means of capturing an image and displaying it on standard video equipment). The frame grabber contains a 7 bit analog to digital converter with fast access memory, accepting video information at a rate of l0 MHz and storing a complete
position adjustment knob (rotationproduces x directiontranslation) y
X
-
~
,.i..~t. modifiS d micro manipulation
y tissue
dish
Volume15, Number 3, 1989
positionsensing transducer tissue mountingblock
Fig. 1. A miniature ultrasonic probe is mounted on a modified micromanipulator for scanning a tissue specimen located in a dish of saline. A U'ansducer is added to detect the X axis movement which becomes the vertical axis on a TV
monitor. The ultrasonic probe sends energy in the Y direction and obtains echoes displayed on the horizontal axis of the TV monitor. ducer to be a part of the fiberoptic endoscope but rather an accessory to it, an important practical advantage. We report the engineering aspects of the system, provide specifications of it and give some examples of images obtained with the new system. More thorough evaluation of its use in vivo is being reported elsewhere. SYSTEM DESCRIPTION The imaging concept of the unit is simple since it functions with linear mechanically scanned probes.
Ultrasonic transceiver
Panasonic
video monitor
Ultrasonic~" ~ - ~
l[ displacement I Displacement transducer
i
Coreco Oculus 200 video frame grabber
Probe I
]E A
v
1
Timing and interface
I
Fig 2. Block diagram of the electronic section of the 20-MHz system which operates with either an in vitro echo probe or an in vivo endoscopic echo probe. The image formation and display output occurs in the video frame grabber. It receives the echo information from the transceiver, timing and control from the timing and interface section, and the computer.
A 20-MHz ultrasound systemfor imagingthe intestinalwall • R. W. MARTINel aL
512 × 480 pixel (7 bits per pixel) image. Further, the computer can operate on this image to modify, magnify, or enhance it via special image processing and manipulation programs provided with the frame grabber (Gray Library, Coreco Inc.) and via some special programs we have developed. This program library consists of approximately 75 subroutines which can be called from the C language. The image captured by the frame grabber is continuously displayed on the black and white monitor (Fig. 2), and any manipulation to it performed by the computer immediately appears in the image. The frame grabber was designed to operate with video cameras, and we have retained this feature as a switch selectable function to use in recording video images of histology obtained with a video microscope. However, for ultrasonic imaging, we modified the frame grabber and linked it to the rest of our system so that ultrasonic echo and ultrasonic transducer position information could be used to generate an image. This refinement required minor modifications to the Coreco board, as well as building the timing and interface circuit indicated in Fig. 2. This timing and interface board (details in Fig. 3) provides the coordination between the frame grabber, the ultrasonic transceiver, and the probe displacement transducer. This displacement transducer senses the position of the probe and is used to insure that each scan line is properly positioned.
The key elements of the method of controlling the frame grabber to acquire, store, and display images are shown in Fig. 3. Minor modifications were made to the Coreco board to allow interrupting the normal control of writing to memory and the memory row addressing. We accomplished the first task by routing the write-enable signal (WR) in the Coreco board to the control and timing of our timing and interface board. We then resubmitted it to the Coreco board as the WRE signal but timed in the manner we required. Next, we replaced the Bus Memory Address Multiplexer chips in the Coreco board with chips similar in function but with outputs that can be disabled. The inputs and outputs of these replaced chips were connected in parallel to identical chips in our timing and interface board. The control and timing circuit in our board disables and enables (BEN1 and BEN2 signals) these Bus Multiplexers to switch the address control between the Coreco board and our timing and interface board. When acquiring an ultrasonic image the addressing of the memory, during the write to memory cycle, is switched to the timing and interface board. There the row addressing is taken from the output of the probe displacement transducer. This displacement signal is converted to a digital number by the analog to digital convener (Fig. 3) and then latched and held during the time each scan line is acquired. This digital number is in the correct format to directly address the memory and force the
CORECO OCULUS 200 BOARD Row Bus and multi )lexor column at ess s __ 512 x 480 x 8 Bit image "--1 ---~ memory
i Control and timing
l
*
WRE
,.u,re image
Probe displacement signal
= Control I and timin!
BEN 1
I
BEN 2 ;
Row address ~
multiplexor
Latch TIMING
275
AND INTERFACE BOARD
Fig. 3. Block diagram of the key elements of our methodology of controlling the frame grabber to acquire, store, and display images are shown. The abbreviations are: WR = write to memory command signal, WRE = write to memory enable, BEN 1 = bus # 1 enable, BEN2 = bus #2 enable, EL = enable latch, and A/D = analog to digital COrlverter.
276
Ultrasound in Medicine and Biology
Volume 15, Number 3, 1989
digitized echo signal to be written to the correct position to generate the proper B mode line. We developed a method to improve the sampling resolution established and limited by the sampling rate of the frame grabber. The restriction is the l0 MHz sampling rate of the analog to digital converter in the Coreco 200 frame grabber. We doubled this rate by using a technique similar to that employed in sampling oscilloscopes. Considering first the method for generating a scan line, the transducer is excited and the echoes resulting from this transmission are digitized and stored in the frame grabber as a partial B mode line. The digitization at the l0 MHz sampling rate is synchronized to the positive phase of the transducer excitation signal. Next, after the acquisition of the first set of echoes is complete, but before the completion of the horizontal sweep, the transmitter is again activated. This time, the beginning of the acquisition is delayed one half of a 10 MHz cycle. This second set of information is stored sequentially as another part of the B mode line. Because of this phase delay, sampling from this acquisition permits sampling between the former sampling points. The net effect is that information is acquired at twice the number of spatial points and two images are acquired this way. The final step is to merge the two B mode images together to form one image. This merger is accomplished by placing each sequentially acquired pixel in the second image after the corre-
[I
same.
Consequently, the transceiver consists of two identical transmitting and receiving channels whose output is multiplexed to a common detector circuit. The envelope of the RF echo data is extracted by this last circuit and provided as the output to the frame grabber. The timing, control and time varying gain control section is dual in nature, in that it provides the transmit enable commands (EN 1 and EN2), the
ULTRASONIC TRANSCEIVER
I Zone 1 I Matching transducer I M1 I
sponding pixel acquired in the first image. The result is an image line with twice the number of points and spatial sampling every 0.035 mm along the radiation path. The ultrasonic transceiver, the second key component of the system, operates with the ultrasonic transducer(s) to acquire echo information from the tissue (Fig. 2). A block diagram of the transceiver appears in Fig. 4. The system has two alternate modes, either single or dual channel. In single mode, one transducer is used to image the entire field of interest ranging 1-20 m m from the transducer. In dual mode, two transducers are used. One element is optimized for imaging a field adjacent to the transducer (1-5 mm); the second, for imaging a field extending from 5-20 mm. Images from both fields are acquired line by line by switching between channels, then combining them to produce a composite image. In the single channel mode the system is locked to one channel, but the rest of the system functions the
Transceiver #1
.....H
Channel T. .I.jG1}'~ multiplexor EN1
I
transducer
I
I Adjust ~ TVG CH1 & CH2 I base and slope gain ! ! I I !
single or dual L channel mode
Detector Ultrasonic echo signal I
M2
_~ ~ I I LI
|
I
.
EN21 I
Timing control and TTme varying gain_ control ('I'VG)
I
I- -
[TVG2 I ~ ~ I
[ I I EN1 - ~ TVG1
I
I I I I ! ! To/From Fr.a~negrabber I Timing and interface
Timing/Synch j
Fig. 4. A block diagram o f the ultrasonic transceiver is shown. The unit consists o f two identical transmitting and receiving channels whose outputs are multiplexed to a c o m m o n detector circuit. The envelope of the R F echo data is extracted by this last circuit and provided as the output to the frame grabber.
A 20-MHzultrasoundsystemfor imagingthe intestinalwall• R. W. MARTINel al. time varying gain control (TVGI and TVG2), and the channel multiplexer control (CHS). PROBE DESCRIPTION We have built eleven 20 MHz probes of various types: eight for in vitro studies and three for endoscopic use. All the probes use gold plated PZT5 piezoelectric material obtained from Valpey-Fisher Corporation, Hopkinton, MA. Miniature disk elements were cut from larger disks using miniature diamond core drills (Lindsay Enterprises Inc., Seattle, WA). The elements are operated in the thickness mode and are backed with air over most of their back face. The diameter of elements ranges between 1.0 and 1.8 ram. Electrical connection is by soldering (Martin et al. 1988), and the front face is coated with a bio-compatible epoxy (Epoxylite 6001, Epoxylite Co., South E1 Monte, CA) after the elements are connected and mounted. In vitro probes are mounted in an assembly that allows easy interchange with other ultrasonic transducers for comparison studies.
/3
The endoscopic imaging probes are designed to be used as an accessory to a fiberoptic endoscope. First, they are built small enough to be passed through the biopsy channel of the endoscope. Second, the probe is constructed so that the tip can be moved along the intestinal wall by sliding the probe in and out of the endoscope. The tip movement is detected by a displacement transducer that is readily attached to the hand piece of the endoscope. Thirdly, the ultrasonic transducer is mounted at the probe tip so it radiates perpendicular to the probe's axis. The signal from the displacement transducer and the echo signals from the ultrasonic transducer as it is moved along the wall allows the system to produce the desired B mode image. Details of the design of the tip are given in Fig. 5. The design allows the ultrasonic element in the tip to be rotated so that the transducer's radiation beam is normal to the intestinal wall. The outer body of the major length of the probe consists of a thin wall Teflon catheter of 2-m length. Inside this catheter is a special cable designed to have high flexibility but
(6w)
(.350)
2m
8.9 mm
/Teflon /tubing mm OD
Glued on I silk I / thread,
/Painted & /coated / w i t h Epoxylite
~
//Needle
ooo 8888o888)888'888
xiglass
~
o
~'/.,"A~/.AI'e'¢,,¢,¢,¢~.~I ~
Triple spring 1.00 mm ID
~l
~
AT 0 0 0 0 0 0 O_0 " ~ ~ 0 0 0 'O00~\".\V'~k\X
\ 'Stainless. \ .~\ \ X \ steelstrain Crimped 50 ohm \a,~,,.~¢~,, re,e) tubing ~wu.~.,,. v..eT___ coax coup,,.ng .46 rnm OD ~lllcone
sheath
277
material
20 mHz'~ Plexiglass 'IgiSm)m xta xtal xtal mount ,.s 1.8 mm OD
3 mm OD
Fig. 5. A cross-sectional drawing of the tip of the endoscopic probe used for in vivo scanning is shown. The 20-MHz piezoelectric clement radiates down in this illustration. An acoustical coupling media is included between the surface of the transducer and the surface of the probe. This coupling provides a delay before the signal encounters the mucosa when the probe is placed directly on the mucosa. The delay prevents the echo signal from being obscured by the ring-down from the excitation pulse. A silicone sheath is used to seal the probe tip from the teflon tubing so that fluids from the bowel will not leak around the inner cable of the catheter. The sheath is flexibleenough to allow the tip to be rotated with respect to the teflon tube over a limited range (two full 360° turns). The inner cable attaches to the tip assembly as illustrated. When the cable is rotated, the tip assembly also rotates. This rotational action allows orientation of the transducer towards the intestinal wall to obtain the optimal cross-sectional image of the wall. The inner coaxial wire connects to the piezoelectric element as shown.
278
Ultrasound in Medicine and Biology
Volume 15, Number 3, 1989
good torsional properties (Martin et al. 1988). This cable is used to transmit rotational force from the proximal end to the piezoelectric element located on the distal tip. Located inside this cable are the electrical wires which connect to the transducer. SPECIFICATIONS OF THE SYSTEM Special computer operations of the system include a program to read in and merge the ultrasonic images, programs to store or recall images from either floppy or hard disk memory and a program to reduce four single full screen images into a composite of four in one screen image. A program also allows outlining regions in images to enable calculation of linear dimensions and areas which can then be compared quantitatively. The band width of the receiver is 5 MHz; the noise referred to the input is 8-tzV rms, and the maximum gain is 85 dB. The time varying gain provides 70 dB of gain control. The receivers optimal noise source impedance (Motchenbacher and Fitchen 1973) is 50 r, as is the transmitter's output impedance. The maximum transmitted voltage into a 50 fl load is 50 Vpp for a 5.5 cycle burst. The typical impedance at 20 MHz of a transducer element connected by 2-m cable of 50 f~ characteristic impedance is 80 + j26 ft. We electrically match the transducers to the system's 50 fl impedance to optimize performance. The electrical to acoustic transfer of one of the endoscopic ultrasonic probes (single element 1.8 mm in diameter) was measured in the following way. A 50 fl output impedance linear power amplifier (RF Communication Inc. #805) was driven by a gated generator (Wavetek Inc. # 164) so it would produce 5 cycles of a 20 MHz sine wave at a repetition rate of 100 kHz. The generator was adjusted so the power amplifier applied 38 Vpp into a 50-fl load. The resistive load was then disconnected and the probe was connected to the power amplifier. The probe was then submerged in water and was directed at the absorber of a Nelson Radiometer #SA-200-002 (NTR Systems, Inc., Seattle, WA). The radiated acoustic power was then measured and the applied electrical power computed. The ratio of electrical to acoustic transfer was then calculated giving a value of -18.6 dB. The 6-dB radiation pattern (Fig. 6) of one of the probes was obtained by imaging seven 0.15-mm diameter nylon strings in a water bath. The image was acquired by first setting the gain appropriately for the experiment. A 3-dB attenuator circuit was introduced
Fig. 6. Image made with the 20-MHz system by scanning 0.15-mm diameter nylon strings is shown. The transducer
was a single disk 1.8 mm in diameter. These strings are suspended in a fluid filled tank, 2 mm apart in both the horizontal and the vertical direction. The vertical spots at the upper fight of the photographs arose from vertically arranged strings. Other artifacts in the image are produced by reflections from the tank walls.
between the transceiver and the transducer at a point were both where impedance matched to each other and to the attenuator. This action produced a 6-dB attenuation of echoes since the transmitted energy passed twice through the network, first during excitation and second when the reflected energy passed into the receiver. The gain of the receiver was then adjusted while scanning until the strings were barely visible in the image. The attenuation was then removed and the strings were scanned and the image recorded. This recorded image represents all the echoes that had amplitudes within 6 dB of the peak point or central part of the beam and therefore represents the 6-dB beamwidth pattern. The image shown (Fig. 6) is from a single disk 1.8 mm in diameter; the narrowest part of its beam occurs at 9 mm from the probe where the lateral resolution is 0.45 mm. The largest lateral beamwidth in the shown imaging field is 0.85 mm. The thickness of the string target was subtracted in making this determination. For a 1-mm diameter element the best lateral resolution measured with the same string target was 0.65 m m and that occurred at a distance of 5 mm from the probe. The average probe range resolution is 0.21 mm.
A 20-MHz ultrasound system for imaging the intestinal wall • R. W. MARTIN el al.
279
IMAGING EXAMPLES Excised intestinal wall tissues have been imaged in vitro with the 20-MHz LEP probe. An example of an in vitro image of canine cecum acquired with our system is shown in Fig. 7. All the wall layers are clearly evident in this image. The histological image shown was taken from approximately the same region. There is good agreement in the layered appearance seen in the ultrasonic and histological images. Of particuiar note in this image is the thickness of the mucosal region. In vivo studies have been conducted in anesthesized dogs. An example of an image acquired of the esophageal wall at the level where the aorta runs adjacent to the esophagus is shown in Fig. 8. The aorta is visible in this image as well as the cross section of an additional vessel. Note that the mucosa appears to blend with the submucosa in Fig. 8. In man, the mucosa of the esophagus is known to be thinner than the mucosa of the stomach (0.3 mm in the esophagus and up to 1.5 m m in the stomach (Weiss and Greep 1977). The mucosa is too thin to appear as a separate layer with our current system.
rn sm
mp s
Fig. 8. Image of the normal canine esophageal wall acquired in vivo with the 20-MHz echo probe system. The vertical axis represents distance into the tissue with the transducer located in the esophagus in the upper part of the image (full scale = 1.8 cm). Horizontal axis is distance along the esophagus (full scale = 2 cm). Note the presence of two vessels adjacent to the wall. EA---excitation artifact, M--mucosa, SM--submucosa, MP--muscularis propria, and S--adventitia, IVW--inner or proximal wall of the blood vessel, OVW~outer or distal wall of the aorta, AO --aorta, and BV--blood vessel.
The dark layer below the bright mucosa and submucosal layer in the image is esophageal muscularis propia, the second bright layer includes the adventitia and the blood vessel wall. The total wall thickness in the esophagus of the dog has been reported to be approximately 4 mm in the cervical region, 2.5 mm in the thoracic region, and 6 mm in the abdominal region (Evans and Christensen 1979). Since the image was taken in the thoracic region the range of wall thickness of 2-3 mm appearing in the image of Fig. 8 agrees with this report. DISCUSSION
ffl
sit3
mp s
Fig. 7. Upper photograph. An ultrasound image taken in vitro of an excised section of canine cecum: vertical axis = depth into the tissue (full scale = 6 mm); horizontal axis = distance along the tissue (full scale = 1.8 cm). The layers seen on the image (from top to bottom) m--mucosa, sm-submucosa, mpumusculads propria, and s--serosa. Lower photograph. Histology of the same specimen taken in close proximity to the region ultrasonically imaged and scaled to approximately the same size. Note: The ultrasonic image was taken approximately 30° oblique to vertical axis in the photograph. It is oriented as shown for ease in comparison to the histological image.
The computer-basod imaging system we have developed is versatile for researching a variety of questions. It is readily applicable to methods of computer processing of ultrasonic images. Computerized comparisons of histology and ultrasonic images are possible if the histological images are acquired through a video camera since they then can be read into the computer via the frame grabber. The radiation pattern found for the 1.8-mm diameter element agrees quite closely with theory in three ways. First, using the usual equation (Wells 1977) for the near-field and far-field transition point (z' = a2f/c where a = radius of a disk transducer, f = frequency, and c = the velocity of propagation of sound) this transition occurs in water for a 1.8-mm diameter transducer at 11 mm. The narrowest part of a one way 3-dB beamwidth for disk transducers exci-
280
Ultrasound in Medicine and Biology
rated with more than 2 cycles (Weyns 1980) occurs closer to the transducer than z'. We found it occurred at 9 mm for the 1.8-mm diameter probe which agrees with Weyns' simulations. Second, the far-field one way 3-dB beamwidth divergence angle may be calculated from the directivity function (Kinsler et al. 1982) and is given by the arc sin(O.318c/fa). This angle then may be used to calculate a transmission and reflection 6-dB beamwidth pattern at various points. For a 1.8-mm element at a distance of 15 mm this gives a width of 0.8 mm which compares closely with the 0.85 mm we measured. Finally, for a transducer that has a wide enough bandwidth to pass an excitation tone burst with an envelope which approximates a window function, the range resolution will be equal to cT/2 where T = the tone burst duration. Our excitation is 5.5 cycles at 20 MHz which when applied to this equation gives 0.2-ram range resolution in water. This is close to what we measured from the string targets after subtracting the thickness of the string from the echoes. The l-ram diameter transducer also agreed fairly close to the theory except for the point at which the narrowest beam occurred. We measured a distance of 5 mm from the transducer whereas theory predicts it should be less than 3.5 mm. However, we cannot be certain of our measurements because the string pattern spacing of 2 mm may have been too coarse to show the pattern properly. The images that we have acquired with the system illustrate several advantages. First, images of normal intestinal wall can be obtained that correlate closely with anatomical layers. Second, the fact that images were acquired with a single element 1.8-mm diameter or less from a nonfocused aperture is important from the viewpoint of building ultrasonic imaging probes that can be passed through the biopsy channels of endoscopes. The high frequency of 20 MHz opens the door to small aperture transducer design although the depth of penetration may be a limiting factor if the wall is thickened. Several future improvements could be made to the system. First, it would be useful to interface the computer more closely to the control of image acquisition, thus providing ease in changing or modifying scanning sequences. Second, the addition of logarithmic compression in the receiver would be advantageous in improving the quality of the image since only time-varying gain compensation was used in the images shown. Finally, designing the system so it could operate over a wide range of ultrasonic frequencies may make it possible to study a wider range of intestinal wall pathologic conditions.
Volume 15, Number 3, 1989
In summary, the system we have developed allows imaging of the intestinal wall both in vitro and in vivo with high resolution. This dual ability is useful in conducting wall imaging research. Although improvements in the system for in vivo imaging are anticipated, it may be possible in its present form for diagnosis of many forms of intestinal wall disease. AcknowledgmentmWe thank and acknowledge Andy Proctor for his work in building probes, Yehuda Sabag in electronic fabrication, and Don Franklin in ultrasonic imaging and preparing the histological image.
REFERENCES Di Magno, E. P.; Regan, P. T.; Clain, J. E.; James, E. M.; Buston, J. L. Human endoscopic ultrasonography. Gastroenterology 83:824-829; 1982. Evans, H. E.; Christensen, G. C. Miller's anatomy of the dog. 2nd ed. Philadelphia: W. B. Saunders Co.; 1979:456. Kimmey, M. B.; Silverstein, F. E.; Haggitt, R. C.; Shuman, W. P.; Mack, L. A.; Rohrmann, C. A.; Moss, A. A.; Franklin, D. W. Cross sectional imaging method: A system to compare ultrasound (US), computed tomograph (CT), and magnetic resonance (MR) with histological findings. Invest. Radiol. 22:227231; 1987. Kimmey, M. B.; Silverstein, F. E.; Martin, R. W. Ultrasound interaction with the intestinal wall. In: Kawati, K., ed. Endoscopic ultrasonography in gastroenterology. New York: IgakuShoin; 1988:35-43. Kinsler, L. E.; Frey, A. R.; Coppens, A. B.; Sanders, J. V. Fundamentals of acoustics. New York: John Wiley & Sons; 1982:180. Martin, R. W.; Silverstein, F. E.; Kimmey, M.; Jiranek, G.; Proctor, A. B mode imaging and Doppler ultrasonic catheters for use with fiber optic endoscopes. Proceedings of the International Society of Optical Engineering: conference on microsensors and catheter based imaging technology. 1988; 904:121-126. Motchenbacher, C. D.; Fitchen, F. C. Low-noise electronic design. New York: John Wiley & Sons; 1973:35-37. Ohmori, S.; Giorgi, F. J.; Starosta, M. An ultrasound imaging system for invasive applications. Proceedings of IEEE: Frontiers of engineering and computing in health care. IEEE CH 1896; 1983:457--460. Silverstein, F. E.; Kimmey, M. B.; Martin, R. W.; Haggitt, R. C.; Moss, A. A.; Mack, L. A.; Franklin, D. W. Ultrasound and the intestinal wall. Scand. J. Gastroenterol. 21(Suppl. 123):34-40; 1986. Silverstein, F. E.; Martin, R. W.; Kimmey, M. B.; Jiranek, G. C.; Franklin, D. W.; Proctor, A. Experimental evaluation of an endoscopic ultrasound probe: in vitro and in vivo canine studies. Gastroenterology [ 1989]. Sivak, M. V. Special methods and techniques in gastroenterologic endoscopy. In: Sivak, M. V., ed. Gastroenterologic endoscopy. Philadelphia: W. B. Saunders Co.; 1987:181-202. Tip, T. L.; Tytgat, G. N. J. Atlas of transintestinal ultrasound. The Netherlands: Mur-Kostvefloren, Aalsmere; 1986. Tytgat, G. N. J.; Tip, T. L., editors. Proceedings of the 4th International Symposium on Endoscopic Ultrasonography. Scan. J. Gastroenterol. 21(Suppl. 123); 1986. Weiss, L.; Greep, R. O. Histology. 4th ed. New York: McGrawHill; 1977:665-668. Wells, P. N. T. Biomedical ultrasonics. London: Academic Press; 1977:28. Weyns, A. Radiation field calculations of pulsed ultrasonic transducers Part 1: planar circular, square and annular transducers. Ultrasonics 18(4):183-188; 1980. Yasuda, K.; Hidekazu, M.; Fujimoto, S.; Nakajima, M.; Kawai, K. The diagnosis of pancreatic cancer by endoscopic ultraoniography. Gastrointest. Endosc. 34(1):1-8; 1988.