There are two main ways in which ultrasound finds application in surgery. The first makes use of the potential of a highly focused beam to produce local tissue destruction, and the second uses mechanical vibrations at ultrasonic frequencies to drive a blade, saw, metal tip or other instrument.
13.5.1 FOCUSED BEAM SURGERY 188.8.131.52 Principles and Techniques
A technique that is to replace a conventional surgical knife should be reproducible and controllable in its ability to destroy tissue, it should be able to affect a sharply defined region only and preferably should be quick and associated with the minimum of blood loss. High-intensity focused ultrasonic beams have most of these qualities. Focal spots about 1-2mm in diameter and 3-4 mm in length may typically be achieved. The tissue ablation technique based on these high-intensity focused beams has come to be known interchangeably as HIFU (high-intensity focused ultrasound) or FUS (focused ultrasound surgery). The principle of this technique is shown in Figure 13.7.
The use of sharply focused beams to produce lesions in organs at depth within the body without damage to intervening tissue was initially investigated in the brain because interest was expressed in the use of these lesions for experimental neuroanatomy. Other organs that have been studied since include the liver, prostate, spinal cord, kidney and eye.
As described in Chapter 2, focusing may be achieved by a variety of methods (see also ter Haar & Hand 1981). The simplest way is to use a spherical, curved shell of piezoelectric material as the transducer. The focus of such a shell lies on the central axis, near the centre of curvature of the bowl. Field distributions for such transducers can be calculated (see Chapter 2; O'Neil 1949; Kossoff 1979). Focal lesions have been produced in the brain tissue of rats and cats using these focused bowl transducers (Warwick & Pond 1968; Robinson & Lele 1972) and in the prostate, liver, breast and kidney in humans (ter Haar 1995).
Although sharply defined heated volumes can be achieved in this fashion, there is no flexibility as to focal distance. Different focal depths may be achieved using a plane transducer in conjunction with a variety of acoustic lenses. These are usually made from materials in which the velocity of sound is greater than that in water, and so concave lenses are needed to
Figure 13.7. Schematic diagram of the principle of focused ultrasound surgery. The focus is placed within the target volume. T|ssue overlying and surrounding the target is undamaged
Figure 13.7. Schematic diagram of the principle of focused ultrasound surgery. The focus is placed within the target volume. T|ssue overlying and surrounding the target is undamaged obtain a converging beam (Chapter 2). The main limitation of such a lens system is absorption in the lens material. Maximum power transmission is obtained when the transducer and lens are separated by a quarter of a wavelength of impedance-matched material. Such transducer/ lens combinations have been used to produce sharply defined lesions in rat and rabbit livers, and in experimental tissues in vitro and in vivo (Linke et al. 1973; Lee et al. 1979; ter Haar et al.
Fry (1958) described the use of four plane transducers, each fronted by a planoconcave spherical lens. The four beams were brought to a coincident focus and the intensity maximised by suitable phasing of the individual beams.
Emerging phased array technology for high-power ultrasound applications has led to the design of transducers that allow electronic beam steering to facilitate the 'painting out' of tissue volumes larger than that of the focal region (Ebbini & Cain 1991; Goss et al. 1996; Daum & Hynynen 1999; Lizzi et al. 1999).
Two types of transducers have been developed for clinical use: trans-rectal devices (for urological applications) and those designed for extracorporeal use. The trans-rectal devices have both an imaging and a therapy transducer incorporated into one unit that can be inserted per rectum for treatment of the prostate. In one device (Foster et al. 1993) the same 4 MHz transducer is used alternately for imaging and therapy, whereas in another (Gelet et al. 1993), imaging is achieved by a retractable 7.5 MHz transducer and therapy undertaken with a 2.25 MHz source.
Vallancien et al. (1993) modified a commercial lithotripter to make an extracorporeal focused surgery device. In this clinical unit, the 1 MHz source consisted of multiple confocal transducer elements and had a focal length of 320 mm. A 3.5 MHz imaging transducer was placed at the centre of the therapy elements. The ultrasound source was placed below the bed in a large water bath and the beam coupled to the target via a waterproof membrane on which the patient lay. A similar coupling geometry has been used by Cline et al. (1992) in which the single-element ultrasound source is designed to be an integral part of a magnetic resonance scanner. In this device the 1.5 MHz spherical bowl source (10.3 cm focal length) lies in the water bath below the scanner couch. A third device (ter Haar et al. 1998) also uses a 1.7 MHz single-element spherical bowl transducer with a focal length of 140 mm. The treatment geometry for this device is such that a small water bag is placed over the patient in good acoustic contact with the skin, and the therapy source is placed into this bag. Imaging is
achieved in this prototype by reproducible interchange of the therapy and diagnostic transducers.
Kohrmann et al. (2002b) have described a handheld device for FUS ablation. In this probe, the ultrasound energy from a 1 MHz cylindrical source is focused using a parabolic reflector. A 3.5 MHz diagnostic probe is positioned in the centre of the cylinder. The handheld probe used to produce haemostasis, developed in Seattle, comprised a spherically concave piezoelectric disc bonded to a solid coupling aluminium cone using a thin layer of epoxy (Vaezy et al. 1997; Brentnall et al. 2001).
Focused ultrasound surgery applicators are also under construction for endoscopic use. Lafon et al. (1998, 2001) have described a device that has been used clinically for the treatment of bile duct carcinoma (Prat et al. 2001). The applicator is constructed around a 2m long flexible metal shaft of diameter 3.8 mm. The active transducer was a plane piezo-ceramic, gilded element (8x2.8 mm) embedded in a brass head 1cm long, backed by an air cavity, operating at 10 MHz. A 12 mm thick polyethylene envelope encases the transducer. When 14Wcm-2 is used for 20 s, a lesion 8 mm long and 3 mm wide can be produced at a depth of 10 mm. A cylindrical phased array for trans-oesophageal use also has been described (Melodelima et al. 2002).
In general, the focal region used in beam surgery is cigar shaped: an ellipsoid of revolution about the central axis of the field. The distribution of pressure at the focus has the form [2J1(x)/x], the width WA of the focal spot being given by (cf. equation (2.5)):
where t0 is the focal length, a is the transducer radius and l is the wavelength in the tissue.
For a non-attenuating medium, diffraction theory predicts that 84% of the energy at the surface of a circular radiator passes through the focal region (Chapter 2; Hueter et al. 1956). However, in tissue where absorption takes place, and/or with other shapes of radiator, this proportion will be reduced.
The precise shape of any lesion produced will depend on the tissue being irradiated. In homogeneous tissue the lesion may be approximately ellipsoidal. If, however, two tissue types are present, one being less absorptive of ultrasound than the other, the lesion shape is less predictable. This would be the case, for example, in the brain where white matter may be damaged selectively, with grey matter and vascular structures being less sensitive (Fry et al. 1955). The vascularity of the tissue also affects the lesion size. Part of the appeal of this technique is that, if sufficient energy is delivered very rapidly, tissue effects are effectively perfusion independent (Billard et al. 1990; Chen et al. 1993a; Hill et al. 1994). Chen et al. (1993a) have suggested that, in order to achieve this, exposures of <3 s should be used. This agrees with the theoretical predictions of Hill et al. (1994).
Geometrically, the ratio of the length to the width of the ellipsoid depends on the solid angle of irradiation. It can be seen from equation (13.1) that, as the frequency is increased, the width of the focal region decreases for a given amount of absorbed energy. For a brief, intense exposure the lesion volume is approximately linearly dependent on the amount of energy absorbed by the tissue (Johnston & Dunn 1976). The optimum frequency for a specific application becomes a compromise between the need to keep the attenuation low in order to allow sufficient energy to reach the target, and the necessity to have sufficient absorption in the target to ensure an adequate temperature rise. Hill (1994) has shown that the optimum frequency in this regard is one that leads to a total attenuation in tissue of the order of 10 dB. Assuming, for illustration, a representative tissue attenuation of 0.7 dB cm-1 MHz-1, then this would imply an optimum frequency of 2.6 MHz for a target depth of 5 cm and of 1.4 MHz for a depth of 10 cm. An extra consideration is that, in general, a smaller focal region is obtained at higher frequencies if all other source geometries remain the same.
There have been various attempts to collate available data pertaining to threshold intensities for lesion production (see, for example, Fry et al. 1970; Lerner et al. 1973; Frizzell et al. 1977; Johnston & Dunn 1981). It has been suggested empirically (and apparently without rigorous assessment of degree of conformity) that, on a log-log plot of intensity as a function of exposure time three collinear regions can be identified. For intensities of >2x 103 Wcm-2 and exposure times of <4x10-2s, cavitation mechanisms are thought to be involved (Fry et al. 1970); for exposure times of >1 s (Lerner et al. 1973) and intensities of <200 Wcm-2 (Frizzel et al. 1977), thermal mechanisms may be responsible. In the region between (shown on Figure 13.8), the mechanism for lesion production is unclear. Cavitation thresholds (as determined by subharmonic emissions) in brain appear to agree with this classification (Gavrilov 1974). A threshold intensity in the region of 30-40 Wcm-2 seems to exist for exposure times of 102-103s (Johnston & Dunn 1981). It has been proposed that the relationship between intensity and exposure time to produce a lesion is:
where c is a weak function of frequency and possibly of the base temperature of the tissue (Dunn et al. 1975). In an attempt to determine the mechanism for lesion development, it has been shown that the threshold curves can be predicted if it is assumed that the relationship between stress and strain in tissue is non-linear and exhibits hysteresis (Johnston & Dunn 1981) and that the sound travels as plane waves in tissue.
Hill et al. (1994) have proposed a general thermal model for lesion development. The basis for this model is the assumption that, for the short, high-intensity exposures used in FUS only those cells whose temperature is raised above a certain hypothetical threshold temperature are killed. This assumption has been the starting point for a number of models (Lizzi & Ostromogilsky 1987) that have given good agreement with measured temperature distributions for specific transducer configurations. The model of Hill et al. (1994) has more general application because it assumes that the focal beam profiles, both axially and laterally, are Gaussian.
It is known from the 'conventional' hyperthermia field that, for the temperature range 42-46°C, the probability of cell death is a function both of exposure temperature and exposure time, given quantitatively by the empirical relationship:
ln(S/S0) = -kt where k is given by the Arrhenius equation k = A exp(E/RgT)
Here, A is a constant, E is the activation energy of the process, Rg is the gas constant and T is the absolute temperature.
Although this relationship has only been tested experimentally for temperatures up to about 50°C, there is reasonable evidence that it will hold good for exposure times down to 1 s (ter Haar 1986). For shorter time scales than this, it is necessary to extrapolate from longer exposures. The relationship suggests that only a fraction of 10-6 of cells would survive an exposure of 60°C for 0.1 s. At this temperature, lesioned tissue takes on a 'cooked' appearance (Clarke & ter Haar 1997). Experimental and theoretical evidence have led to a general consensus that the temperature at the edge of the FUS lesion lies between 55°C and 60°C (Robinson & Lele 1972; Hill & ter Haar 1995).
Figure 13.8. Plot of log(Intensity) against log(T|me) to show the threshold for lesion production using focused beams. The line has the form It1/2=c(f,7) (see text). The domains for thermal and cavitational mechanisms of damage are shown log Time (sec)
Figure 13.8. Plot of log(Intensity) against log(T|me) to show the threshold for lesion production using focused beams. The line has the form It1/2=c(f,7) (see text). The domains for thermal and cavitational mechanisms of damage are shown
The Arrhenius relationship and the assumption of Gaussian beam profiles are used by Hill et al. (1994) to derive a 'lesioning rate parameter', R, which is the reciprocal of the time required to achieve a lesion threshold in the absence of thermal redistribution. Theory and experiment have been shown to be in good agreement at low-intensity exposures (see Figure 13.9). The deviation of experimental measurement from theoretical prediction at the higher intensities is probably a result of non-linear and cavitation effects.
Figure 13.10 shows two lesions. The one on the left is a clinically 'useful' lesion, ellipsoidal in shape and falling symmetrically across the focal plane. It is made using an exposure close to the threshold. The lesion on the right is misshapen and is the result of either too high an intensity or too long an exposure time. In a detailed study of the effect of increasing intensity and exposure time, Watkin et al. (1996a) showed that, at intensity/time combinations above the threshold for lesion production, the lesion moves forward through the focus towards the source and forms a 'bulbous'-shaped head. Meaney et al. (1998, 2000) have modelled this phenomenon theoretically and have shown that it can, in part, be explained by the increase in acoustic absorption coefficient due to the raised ambient temperature in the beam path. These effects must be taken into consideration when attempting to 'paint out' large tissue volumes by placing single lesions side by side in arrays (Malcolm & ter Haar 1996). In order to avoid effects due to a gradual rise in ambient tissue temperature, it is necessary to wait up to 60 s between successive exposures. An alternative method of delivering the required energy is to move the transducer while it is excited, thus forming scanned 'tracks' as shown in Figure 13.13b. Typical scan rates are 1-4mm s_1.
Was this article helpful?