Using the Acoustic Interference Pattern to Locate the Focus of a High-Intensity Focused Ultrasound (HIFU) Transducer

Ultrasound in Med. & Biol., Vol. 34, No. 1, pp. 137–146, 2008 Copyright © 2007 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/08/$–see front matter

doi:10.1016/j.ultrasmedbio.2007.07.001

● Original Contribution USING THE ACOUSTIC INTERFERENCE PATTERN TO LOCATE THE FOCUS OF A HIGH-INTENSITY FOCUSED ULTRASOUND (HIFU) TRANSDUCER CHIH-CHING WU,*† CHIUNG-NIEN CHEN,‡¶储 MING-CHIH HO,‡¶ WEN-SHIANG CHEN,†§¶ ‡¶ AND PO-HUANG LEE *Department of Mechanical Engineering, National Taiwan University, Taipei, Taiwan; †Division of Medical Engineering Research, National Health Research Institute, Zhunan, Miaoli, Taiwan; ‡Department of Surgery, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan; § Department of Physical Medicine and Rehabilitation, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan; ¶Angiogenesis Research Center, National Taiwan University, Taipei, Taiwan; and 储Division of Mechanics, Research Center for Applied Sciences, Academia Sinica, Taipei, Taiwan (Received 25 October 2006; revised 4 June 2007; in final form 2 July 2007)

Abstract—One of the main problems encountered when using conventional B-mode ultrasound (US) for targeting and monitoring purposes during ablation therapies employing high-intensity focused US (HIFU) is the appearance of strong interference in the obtained diagnostic US images. In this study, instead of avoiding the interference noise, we demonstrate how we used it to locate the focus of the HIFU transducer in both in vitro tissue-mimicking phantoms and an ex vivo tissue block. We found that when the B-mode image plane coincided with the HIFU focal plane, the interference noise was maximally converged and enhanced compared with the off-focus situations. Stronger interference noise was recorded when the angle (␣) between the US image plane and the HIFU axis was less than or equal to 90°. By intentionally creating a target (group of bubbles) at the 3.5-MHz HIFU focus (7.1 mm in length and 0.7 mm in diameter), the position of the maximal noise convergence coincided well with the target. The differenced between the predicted focus and the actual one (bubbles) on x and z axes (axes perpendicular to the HIFU central axis, Fig. 1) were both about 0.9 mm. For y axis (HIFU central axis), the precision was within 1.0 mm. For tissue block ablation, the interference noise concentrated at the position of maximal heating of the HIFU-induced lesions. The proposed method can also be used to predict the position of the HIFU focus by using a low intensity output scheme before permanent changes in the target tissue were made. (E-mail: [email protected]) © 2007 World Federation for Ultrasound in Medicine & Biology. Key Words: High-intensity focused ultrasound, Interference, Ablation.

INTRODUCTION

netic resonance imaging), fast acquisition speed and ability of dynamic examination. A successful treatment requires reliable US images for precise targeting and realtime monitoring so as to ensure that the HIFU energy is delivered to the target tissue while avoiding damage to the adjacent healthy structures. However, the high-intensity acoustic waves from the HIFU transducer interfere with the US images, sometimes completely saturating the receiver electronics of the diagnostic US system, resulting in bright echoes throughout the image. This is especially problematic when HIFU is performed in continuous mode. Previous studies have employed various methods to synchronize HIFU excitation with diagnostic US imaging to allow the treatment site of HIFU to be visualized in real time (Owen et al. 2006; Vaezy et al. 2001a). The

High-intensity focused ultrasound (HIFU) is a promising tool in applications such as tumor ablation, vessel destruction and bleeding control due to its noninvasive nature (Bailey et al. 2001; Jolesz et al. 2005; Roberts, 2005; Vaezy et al. 2001a; Watkin et al. 1996; Wu et al. 2002, 2004). Various commercialized HIFU systems use diagnostic ultrasound (US) for targeting and monitoring, partly due to the low cost of US (compared with mag-

Address correspondence to: Wen-Shiang Chen, MD, PhD, Department of Physical Medicine and Rehabilitation, National Taiwan University Hospital, No. 7, Zhongshan S. Road, Taipei 100, Taiwan. E-mail: [email protected] 137

138

Ultrasound in Medicine and Biology

Volume 34, Number 1, 2008

Fig. 1. (a) Experimental set-up and (b) definition of ␣ and ␤ angles. The function generator and display system were not synchronized. The white dashed circle in (c) enclosed the central spindle area of the interference noise, whose center (white arrow) was assumed the focus of the HIFU transducer on x-z plane. The width of the central spindle (d ⫻ 1) over the width of noise 10 mm below (d ⫻ 2) was defined the convergence ratio (d ⫻ 1/d ⫻ 2).

synchronization ensures that the interference noise is stable at a particular section of a US image, allowing visualization of the treatment site through a noise-free window. The HIFU treatment results in the generation of hyperechoic spots at the focus, which reportedly have a threshold intensity below that required for immediate cellular damage (Khokhlova et al. 2006; Vaezy et al. 2001b), thus, enabling pretreatment targeting. However, the threshold for generating hyperechoic spots differs with the tissue type and is unknown without preliminary testing. Moreover, clear visualization requires interruption of the HIFU treatment, resulting in decreased output energy and increased ablation time. Ultrasonic estimation of the temperature-induced “echo strain” has also been suggested as a noninvasive technique for guiding HIFU therapy and predicting the location of the thermal lesion before it is formed. Miller et al. (2004, 2005) showed that ultrasonic temperature imaging is able to resolve a temperature rise of less than 2°C and thus is able to target the HIFU focus noninvasively. However, the temperature mapping requires radiofrequency (RF) data to be obtained and modified signal-processing algorithms, which are not available in current commercial US systems and the image quality is affected by artifacts from the thermoacoustic effect.

In this study, we demonstrate a new and easy method for determining the HIFU focus using US images based on the interference pattern itself. MATERIALS AND METHODS Experimental set-up Figure 1 shows a schematic of the experimental setup. A 3.5-MHz single-element focused piezoelectric transducer (Sonic Concepts, Woodinville, WA, USA) was mounted on one side of a transparent tissue-mimicking phantom orthogonal to a diagnostic US transducer (L38, SonoSite, Bothell, WA, USA). The L38 transducer is a 5 to 10 MHz broadband linear array transducer with a centre frequency of 7.5 MHz. With such an arrangement, an x-z slice of the phantom could be visualized by diagnostic US. Angle ␣ in the figure represents the tilt angle of the diagnostic probe relative to the surface of the phantom, while angle ␤ is the rotation angle around the central axis of the diagnostic probe (Fig. 1b). The temperature-sensitive transparent tissue-mimicking phantoms used in the current study were constructed from acrylamide/bis gel and albumin as reported previously, which turned white while its temperature exceeded 55°C (Lafon et al. 2005; Takegami et al. 2004).

Ultrasound targeting using HIFU noise ● C.-C. WU et al.

The HIFU transducer we used has an aperture diameter and a radius of curvature of 35 and 55 mm, respectively. The focus depth measured from the plane crossing the outer rim of the HIFU transducer’s front surface in the output direction is about 50 mm. The focus width (x- and z-directions) and length (y-direction), as measured by a needle hydrophone (SPRH_S_0500, ONDA, Sunnyvale, CA, USA), were 0.7 and 7.05 mm, respectively. The incident and reflected electric powers were measured using a power meter (4421, Bird, Cleveland, OH, USA). Assuming the efficiency of the transducer is 50%, the spatial-average–temporal-average intensity (ISATA), calculated as the average acoustic power over the ⫺3 dB cross-sectional area of the focus, were 5.4 and 288.7 W/cm2 for electrical powers of 1.5 W (for targeting) and 80 W (for producing lesion) in water, respectively. The HIFU and display systems did not need to be synchronized, but a continuous-wave output from the HIFU transducer was necessary. All experiments were performed in a 24 ⫻ 21 ⫻ 15 cm acrylic tank containing degassed water at 37°C. The degassed water was prepared by boiling tap water and then cooling to 37°C in a sealed container. The dissolved oxygen level could be reduced to less than 4 ppm in this manner. The RF signal was supplied by a function generator (33120A, Agilent, Palo Alto, CA, USA) whose output was amplified by an RF power amplifier (150A250, Amplifier Research, Souderton, PA, USA). The incident and reflected powers were measured using a power meter (4421, Bird, Cleveland, OH, USA). Experimental series Series of experiments was performed to demonstrate how the interference noise pattern of HIFU on diagnostic US images can be used to locate the focus of HIFU treatment. Experiment 1: noise pattern and output intensity of HIFU. An acrylic tank with a HIFU transducer mounted on its side wall was first filled with partially degassed tap water. A tissue-mimicking phantom was then fixed in the water tank at the position where the HIFU focus was near the center of it. HIFU waves of different electrical powers ranging from 0.05 W to 1.5 W (for focus localization) and 80 W (for producing lesion) were applied and the resultant noise patterns on B-mode images were recorded and compared. The B-mode imaging plane was perpendicular to the HIFU axis and crossed the focal region. Experiment 2: interference noise pattern at the focus. To obtain the information of the interference pattern at the HIFU focus, the tissue-mimicking phantom was used again in this part of experiments and was carefully scanned along its upper surface to obtain planes of images perpendicular to the HIFU axis (y-axis). US

139

images at distances between 0 and 60 mm (focus at 50 mm) from the HIFU exit planes (the plane crossing the outer rim of the HIFU transducer’s front surface in the output direction) were recorded every 0.5 mm while the HIFU transducer was driven at a low electrical power level (1.5 W) in continuous-wave mode. The diagnostic probe was also moved from left and right by a stepper motor to ensure that the interference noise remained in the center of the images. The angle between the HIFU probe and the surface of the tissue-mimicking phantom (angle ␣) was manually set at 90 ⫾ 5°. The interference noise appeared as a hyperechoic spindle at the center with periodical hyper- and hypoechic bands around. The center of the spindle was assumed the focus of the HIFU transducer, and the width of the spindle (dx1) over the distance between the outer borders of the first “side lobes” of interference noise 1 cm below bilaterally was defined as the convergence ratio (dx1/dx2) (Fig. 1c). Both the maximal signal intensity at the predicted focus and the convergence ratio of the noise were evaluated. Experiment 3: determining the accuracy of focus prediction. Bubbles could be easily produced at the focus of a HIFU transducer, assuming the position of maximal heat deposition, by strong and short ultrasonic waves through boiling (Khokhlova et al. 2006; Tung et al. 2006). To evaluate the accuracy of focus prediction, bubbles were deliberately produced in a phantom as a marker of the HIFU focus by powering the HIFU transducer at 80 W for 1 s. To observe the noise pattern and compare the locations of bubbles and maximal noise convergence, the HIFU transducer was then switched to the low power mode (1.5 W). Phantoms were carefully scanned along the y axis in 1 mm interval. The discrepancy between the predicted focus and the position of hyperechoic bubbles were also measured on x-z plane. Moreover, to evaluate the effect of tilting and rotation angles (␣ and ␤ angles) of the diagnostic probe on the interference patterns, the noise patterns of four angle combinations, ␣ ⫽ 85°/␤ ⫽ 0°, ␣ ⫽ 45°/␤ ⫽ 45°, ␣ ⫽ 60°/␤ ⫽ 0°, and ␣ ⫽ 95°/␤ ⫽ 0° were recorded and shown as examples. Experiment 4: Testing on ex vivo tissue blocks. It is critically important to see if the predicted focus coincided with the maximally heating position in real tissue, especially the accuracy along the z axis. The HIFU produced lesion was cigar or tadpole shaped and extended more than 2 cm along the y axis (Watkin et al. 1996) and, thus, the accuracy of prediction was not critical. For x axis, the noise converged well which left no much doubt about determining the location of focus. However, since the noise extended in both directions and was not necessarily symmetric along the z axis, it was important to determine if the maximally heated position

140

Ultrasound in Medicine and Biology

Volume 34, Number 1, 2008

Fig. 2. (a) The maximal intensity of the interference noise on B-mode US images for HIFU output (in electrical power) between 0.05 and 1.5 W; (b) Examples of interference noise at 0.05, 0.1, 0.5 and 1.0 W (tissue mimicking phantoms); (c) HIFU-induced interference noise at the focal plane for high (80 W) and low (1.5 W) electric power conditions (pig liver) and (d) the signal intensities of a horizontal line (dashed lines) crossing the predicted focus of (c).

Ultrasound targeting using HIFU noise ● C.-C. WU et al.

141

for example) because of a better background contrast (Fig. 2c and d). As shown in Fig. 2d, the signal peak near the predicted focus at 1.5 W is 75 to 100 levels of gray-scale brighter than the “valley” of signal around. However, at 80 W, several peaks of comparable intensities emerged, which made the determination of the focus position difficult.

Fig. 2. (Continued)

coincided with the predicted focus. By replacing the acrylamide phantom with tissue blocks of either porcine thigh muscle or liver, the interference noise produced by the same HIFU transducer operating at a relatively low power (1.5 W, determined by experiment 1) was used to locate the HIFU focus inside tissue blocks before the occurrence of permanent changes. The power of the HIFU transducer was then increased to 30 W for 20 s, after which the tested blocks were cut open to compare the position of maximal color change with the predicted focus position. The position of the maximal color change in tissue block was assumed the position of maximal tissue necrosis and also the position of maximal heat deposition. RESULTS The intense acoustic waves from a HIFU transducer formed a special pattern of noise for image planes perpendicular to the HIFU axis. Near the HIFU focus, a vertical spindle-shaped noise could be easily seen on the display of the B-mode ultrasound (Fig. 1c). The smaller the convergence ratio was, the better the noise converged, and the easier for the observer to determine where the focus was on the x-z plane. Experiment 1 As shown in Fig. 2a, the intensity of the interference noise increased with the output of the HIFU transducer, but the increase was less prominent after about 0.5 W. Moreover, since the noise outside the focus area also increased, the signal to noise ratios did not increase as well. Selected examples were shown in Fig. 2b. Comparing the interference noise before and during the HIFU ablation, the noise at the HIFU focus was actually more evident at low output levels (for electric power of 1.5 W,

Experiment 2 The images obtained in experiment 2 are shown in Fig. 3. a to f. When the image plane (x-z plane) coincided with the focal plane (50 mm, Fig. 3e), the interference noise was maximally converged (Fig. 3h) and enhanced (Fig. 3g). The actual focus of the HIFU transducer on the z-axis approximately coincided with where the convergence of the interference noise was the narrowest. No interference noise was seen when the HIFU transducer was operated in pulsed-wave mode (results not shown). Interference noise at other distances along the HIFU axis (i.e., y-direction) was sometimes evident (peaks in Fig. 3g except focus). However, the noise at these distances was either less converged (Fig. 3h) or weaker in intensity (Fig. 3g). For x direction, convergence noise is clearly evident at the center of the images shown in Fig. 3e, which provides a good estimate of the focus position along the x-direction. However, in the vertical direction (z direction), the interference noise shows as a spindle-shaped hyperecho extending for almost 2 cm, which sometimes made it difficult to determine the exact position of the focus. We assumed that the center of this spindle-shaped noise pattern was the HIFU focus. Experiment 3 The bubbles created by short and high-intensity acoustic waves from the HIFU transducer appeared as bright hyperechoes at the focus of the imaging plane. After switching the HIFU transducer to a low-intensity mode, clear interference noise was evident at these bubbles on the B-mode images (Fig. 4). Before creating the marker bubbles, the maximal intensity noise was found at 50 mm (Fig. 4e) as expected from information of beam plotting using needle hydrophone. Bubbles at the HIFU focus drastically increased the intensity of the interference noise (Fig. 4e and Fig. 5). Along the y axis, the bright echo producing by the created bubbles also the maximal values at 50 mm as predicted (resolution along the y axis was 1 mm). However, the maximal intensity position along the x axis shifted right from pixel 123 to 129, as shown on Fig. 5c, or about 0.93 mm. For z axis, the shift was about 0.92 mm. If the center of the spindleshaped noise in Fig. 5a was assumed the focus, the center of the bright noise (the bubbles in Fig. 5b) shifted approximate 10 pixels on the x-z plane or about 1.40 mm.

142

Ultrasound in Medicine and Biology

Volume 34, Number 1, 2008

Fig. 3. B-mode US images containing interference produced by HIFU in an acrylamide tissue-mimicking phantom or water (x-z planes) at distances of (a) 10, (b) 20, (c) 30, (d) 40, (e) 50 (focus) and (f) 60 mm along the y axis. The distance 0 was set at the plane incorporating the front outer rim of the HIFU transducer (the HIFU exit plane). The vertical direction of the images corresponds to the z-axis shown in Fig. 1, while the horizontal direction is the x-axis. The maximal signal intensity (g) and convergence ratio (h) along the y axis are also shown.

As shown in Fig. 5b, the presence of bubbles, a strong scattering region, also changed the interference pattern and moved the spindle-shaped noise on Fig. 5a upward (in z direction). When the ␣ angle was close to 90° (Fig. 4a and d), the interference noise approximately exhibited a symmetrical pattern to z axis on the 2D B-mode images. However, when the imaging probe was rotated (␤ ⫽ 45°) as shown in Fig. 4b, the interference lost its symmetry. Experiment 4 To determine the accuracy of prediction along the z axis for ex vivo conditions, the maximally converged center of the interference noise formed at a low power mode (the presumed focus of HIFU transducer on x-z

plane, Fig. 6a) was moved by a step motor to a point 16 mm below the upper surface of the porcine thigh muscle (the target). The HIFU transducer was then switched to the high power mode for tissue ablation. After a single ablation, the tissue blocks were sliced orthogonal to the imaging plane to reveal both the whole lesion along the y axis and the distance between the surface and the lesion center. The HIFU waves came from the right side in Fig. 6b, resulting in a tadpole-shaped lesion with its head toward the right side. By careful measurement, the central axis of the formed lesion was found to situate at the same position along the z axis, 16 mm, as expected. Similar results were obtained in porcine liver blocks, as shown in Fig. 6c and d. Both the maximally converged center of the interference noise before ablation and the

Ultrasound targeting using HIFU noise ● C.-C. WU et al.

143

Fig. 4. The interference noise for four values of angle ␣: (a) 85°, (b) 45° (rotating 45° first or ␤ ⫽ 45°), (c) 60° and (d) 95° and (e) along the y-axis before and after creating bubbles at the HIFU focus.

axis of the tadpole-shaped lesion were situated at a distance 6 mm below the upper surface of the tissue block. DISCUSSION AND CONCLUSIONS This study demonstrates that the interference noise of HIFU on diagnostic ultrasonic images can be used to locate the actual focus of the HIFU transducer during ablation. From experiment 1, since only a low intensity HIFU is needed to produce a visible noise pattern, it is possible to predict the focus position without inducing any permanent tissue changes (e.g., protein denaturing) or bubble formation, which may change the lesion shape and move the maximally heating position along the y axis (Tung et al. 2006). The position of the imaging probe to the axis of the HIFU transducer is important for the formation of inter-

ference noise. Good convergent pattern could be easily obtained along the y axis but away from the focus, which increased the difficulty of focus localization. Since noise converged the best at the focus, a thorough scan along the y axis might be necessary to differentiate a better and the best convergent positions. In experiment 3, we recorded B-mode images in a resolution of 1 mm along the y axis. The predicted focal position before creating marker bubbles matched with the position of the hyperechoic bubbles, the presumed focal position (50 mm from the exit plane). Thus, the accuracy along the y axis was less than 1.0 mm. After determining the focal position along the y axis, the position of the focus on x-z plane could then be determined. Our experimental results showed that the discrepancy between the predicted focus and the actual one was also about 1.4 mm on the x-z plane.

144

Ultrasound in Medicine and Biology

Fig. 5. The interference noise on B-mode images (a) before and (b) after creating bubbles at the HIFU focus is compared. The pixel brightness of the white dashed line on (a) and also the same location on (b) (crossing the bubbles) is shown on (c).

Angle ␣ is an important variable for maximizing the clarity of the interference noise. Our preliminary testing revealed that the interference noise became more prominent and converged as the angle was made smaller than 90°. Therefore, a large access window is necessary for both the therapeutic and diagnostic probes, which limits the applicability of the proposed method. This may make the proposed method more suitable in applications (e.g., breast tumor ablation) where a larger window is possible. Angle ␤ altered the symmetry of the interference pattern as the example shown in Fig. 4b. Better positioning the imaging probe improves the symmetry of the interference noise and make the judgment of the focus easier. In our phantom studies, no protein denature was found during the localizing process lasting for more than 5 min at the electrical power of 1.5 W (intensity ⫽ 5.4 W/cm2). Therefore, 1.5 W was considered safe and was used through our experiments. However, tissue damage for in vivo condition may still occur because the attenu-

Volume 34, Number 1, 2008

ation coefficient of tissue is higher than the tissue-mimicking phantoms used in our study (Lafon et al. 2005). The maximal intensity for the ultrasonic devices used in physical therapy is usually around 2 W/cm2, at which heat and pain are usually perceived if the transducer is kept stationary for longer than 60 s. Fortunately, the duration of time necessary for targeting is usually short and interference noise is still evident for a power output less than 0.5 W (1.8 W/cm2) in our study (Fig. 2b). Therefore, tissue damage for in vivo condition could be avoidable. The resolution was best in the x-direction, where the noise converged at a region with a width of a few mm (the width of the spindle-shaped noise). For the y-direction (i.e., the HIFU axial direction), the interference noise pattern was evident in a few imaging planes other than the focal plane, although the noise was either less converged or lower in intensity. However, since the length of the lesion was longer than 2 cm along the y axis, the resolution of the prediction along y axis is not critically important. In the z-direction (vertical direction), a short section of hyperechoes (a few mm to more than 1 cm) was seen, which increased the difficulty of determining the exact focus position. We assumed the center of the spindle the focus, and showed that the accuracy was within 1.0 mm. The deliberately created bubbles at the HIFU focus in experiment 2 coincided pretty well with the presumed focus, with a difference of about 1.4 mm on x-z plane. Moreover, the noise was also enhanced, which was probably due to the presence of more scatters (bubbles) at the focus (Fig. 5). Other factors affecting the scattering condition appear to influence the formation of interference noise. The position of the diagnostic US probe, the presence of inhomogeneity in phantoms, and the types of materials under investigation (e.g., phantoms or tissue blocks) may alter the scattering properties of the target region and, thus, the brightness and shape of the interference noise. Therefore, to allow correct conclusions to be drawn from reliable images, it is necessary to move the diagnostic probe around so as to find a better and decisive pattern of noise. The formation of the interference noise is probably attributable to enhanced scattering. Scattering occurs when sound waves propagating through a material meet a region of inhomogeneity, where some of the incident energy is refracted and reflected. The acoustic energy is maximal at the HIFU focus, so the interference noise in the B-mode images is maximal here (Fig. 7). The changes of noise pattern after introducing bubbles in Fig. 5b further suggested the importance of scattering effect in the production of interference noise. The angle between the diagnostic probe and the HIFU axis is also important to the noise intensity. When the diagnostic

Ultrasound targeting using HIFU noise ● C.-C. WU et al.

(a)

145

(b) 16 mm 16 mm

(c)

6 mm

(d) 6 mm

z axis y axis

Fig. 6. Ex vivo ablation guided by HIFU-induced interference noise for (a) and (b) a porcine thigh muscle block and (c) and (d) a porcine liver block. (a) and (c) are B-mode ultrasonic images showing the interference noise and (b) and (d) show the shape and dimension of the formed lesions after cutting open the tissue blocks. The tissue blocks were sliced orthogonal to the imaging plane to reveal the whole lesion along the y axis.

probe is set at a position where the probe surface faces the HIFU transducer (␣ ⬍ 90°) during image acquisition, more scattering sound waves from the target region can be received, and thus the noise is brighter. In experiment 2, the interference noise increased in intensity after intentionally inducing bubbles (strong scatters) at the focus, which further suggested the theory of enhanced scattering. The emitted US pulse from the imaging probe lasts approximately 1 ␮s. It then listens for approximately 99 ␮s. In contrast, the HIFU transducer emits continuously, hence the extended lines of interference is formed along the z axis. The frequency of the HIFU transducer and the bandwidth of the imaging probe both appear to affect the formation of interference noise. One of the reasons for selecting the 5–10 MHz imaging probe and the 3.5 MHz HIFU transducer is their overlapping frequency range. Preliminary testing showed that using a HIFU transducer of lower frequency or an imaging probe of higher frequency both reduced the intensity of the interference noise. The time-gain compensation on the ultrasonic imager should also be adjusted to maximize the clarity of the noise pattern. To maximize the clarity of the noise pattern for clinical practice, a modified imaging probe which covered the main frequency of the HIFU transducer should be used. We have also tested the noise pattern on other available diagnostic ultrasound system,

such as Toshiba Xario equipped with a 6 MHz linear probe operating at 5.8 MHz. One example of the obtained images was shown in Fig. 8. For free-hand scanning using diagnostic probe, an automatic tracking program might be necessary to record and compare the convergence of the interference noise on different imaging planes and suggest the actual location of the HIFU focus. If three-dimensional volumetric scan is available, it may provide an overall view of the noise pattern and shorten the time for determining the focal position. After coinciding the HIFU focus with

Fig. 7. Possible mechanism for the formation of interference noise.

146

Ultrasound in Medicine and Biology

Volume 34, Number 1, 2008

Fig. 8. Interference noise observed by the 6-MHz linear probe of Toshiba Xairo system.

target using low-intensity output from HIFU transducer or the pretreatment targeting, ablation can be initiated and tumor may be destroyed by filling the full volume of a target and a safety zone around with HIFU-induced lesions. Acknowledgments—This research was supported by grants from the National Science Council (No. 94-2622-E-002-020-CC3) and the National Taiwan University Hospital (No. 94A14).

REFERENCES Bailey MR, Couret LN, Sapozhnikov OA, Khokhlova VA, ter Haar G, Vaezy S, Shi X, Martin R, Crum LA. Use of overpressure to assess the role of bubbles in focused ultrasound lesion shape in vitro. Ultrasound Med Biol 2001;27:695–708. Jolesz FA, Hynynen K, McDannold N, Tempany C. MR imaging-controlled focused ultrasound ablation: A noninvasive image-guided surgery. Magn Reson Imaging Clin N Am 2005;13:545–560. Khokhlova VA, Bailey MR, Reed JA, Cunitz BW, Kaczkowski PJ, Crum LA. Effects of nonlinear propagation, cavitation, and boiling in lesion formation by high intensity focused ultrasound in a gel phantom. J Acoust Soc Am 2006;119:1834 –1848. Lafon C, Zderic V, Noble ML, Yuen JC, Kaczkowski PJ, Sapozhnikov OA, Chavrier F, Crum LA, Vaezy S. Gel phantom for use in high-intensity focused ultrasound dosimetry. Ultrasound Med Biol 2005;31:1383–1389. Miller NR, Bamber JC, ter Haar GR. Imaging of temperature-induced echo strain: preliminary in vitro study to assess feasibility for guiding focused ultrasound surgery. Ultrasound Med Biol 2004;30: 345–356. Miller NR, Bograchev KM, Bamber JC. Ultrasonic temperature imaging for guiding focused ultrasound surgery: Effect of angle between

imaging beam and therapy beam. Ultrasound Med Biol 2005; 31:401– 413. Owen NR, Bailey MR, Hossack J, Crum LA. A method to synchronize high-intensity, focused ultrasound with an arbitrary ultrasound imager. IEEE Trans Ultrason Ferroelectr Freq Control 2006;53:645– 650. Roberts WW. Focused ultrasound ablation of renal and prostate cancer: Current technology and future directions. Urol Oncol 2005;23:367–371. Takegami K, Kaneko Y, Watanabe T, Maruyama T, Matsumoto Y, Nagawa H. Polyacrylamide gel containing egg white as new model for irradiation experiments using focused ultrasound. Ultrasound Med Biol 2004;30:1419 –1422. Tung YS, Liu HL, Wu CC, Ju KC, Chen WS, Lin WL. Contrast-agentenhanced ultrasound thermal ablation. Ultrasound Med Biol 2006; 32:1103–1110. Vaezy S, Andrew M, Kaczkowski P, Crum L. Image-guided acoustic therapy. Annu Rev Biomed Eng 2001a;3:375–390. Vaezy S, Shi X, Martin RW, Chi E, Nelson PI, Bailey MR, Crum LA. Real-time visualization of high-intensity focused ultrasound treatment using ultrasound imaging. Ultrasound Med Biol 2001b;27: 33– 42. Watkin NA, ter Haar GR, Rivens I. The intensity dependence of the site of maximal energy deposition in focused ultrasound surgery. Ultrasound Med Biol 1996;22:483– 491. Wu F, Chen WZ, Bai J, Zou JZ, Wang ZL, Zhu H, Wang ZB. Tumor vessel destruction resulting from high-intensity focused ultrasound in patients with solid malignancies. Ultrasound Med Biol 2002;28: 535–542. Wu F, Wang ZB, Chen WZ, Wang W, Gui Y, Zhang M, Zheng G, Zhou Y, Xu G, Li M, Zhang C, Ye H, Feng R. Extracorporeal high intensity focused ultrasound ablation in the treatment of 1038 patients with solid carcinomas in China: An overview. Ultrason Sonochem 2004;11:149 –154.