SCIENCE & MEDICINE

September/October 1996 · Volume 3, Number 5

Copyright © 1996 by Science & Medicine, Inc. All Rights Reserved.

Focused Ultrasound Surgery Guided by MRI

By Kullervo Hynynen

Because it is sensitive to temperature changes in tissue, MRI is an effective method for guiding and controlling ultrasound pulses. Ultrasound beams heat tissue, but the objective of focused ultrasound therapy is tissue destruction, more like surgery than hyperthermia, so it might be tumors in organs accessible to ultrasound or for coagulating blood vessels. A prototype designed for breast tumor surgery is now being tested clinically. Multiple ultrasound transducers in a phased array would permit focused ultrasound surgery to be used for sizable tumors.

One of the dreams of medicine has been to be able to perform surgical procedures without opening the patient. Non-invasive surgery would eliminate scar formation, blood loss, and infections, and it would reduce the risk of other complications. Recovery time would be shortened, and many procedures could be done on an outpatient basis. The absence of tissue penetration would ease requirements for sterility in the operating room. The benefits of non-invasive surgery would also result in a significant reduction in cost.

During the past 20 years this dream has come closer to reality because of the development of imaging techniques that allow tumors and other anatomical structures to be visualized with precision. The first successful application of non-invasive surgery was the use of shock waves to disintegrate kidney stones. In this procedure, stones are located by x-ray imaging and multiple shock waves are aimed at them through the skin and overlying tissues. A stone is slowly broken into small fragments by the energy pulses, and the treatment is continued until the x-rays verify that the stone has been destroyed.

This is an example of a completely noninvasive method that has revolutionized kidney stone treatment and has resulted in the sorts of benefits already mentioned. Such as system cannot be used to destroy soft tissues, but similar concepts have been explored for tumor therapy. Recent advances have shown that such devices may be feasible and that noninvasive surgery of soft tissues may become common practice within a few years.

One completely noninvasive method proposed for deep tissue destruction is focused ultrasound. The sue of ultrasound beams as surgical tools was first proposed more than 50 years ago for destruction of brain tissue. A complex sonication system that used x-rays to determine the target location with respect to the skull was developed by Bill and Frank Fry at the University of Illinois during the 1950's and was tested in treatment of Parkinson's disease, but it was never used outside the research setting.

The principle difficulty at the time was the accurate location of the target tissues. More recently, focused ultrasound surgery systems have been combined with diagnostic ultrasound imaging to make soft tissue tumor sonication possible. Several clinical trials for the treatment of benign and malignant tumors of the prostate, bladder, kidney, and eye have been conducted with these devices.

Although targeting with diagnostic ultrasound works well in many cases, the treatment component still relies on open-loop energy delivery. Power settings are based on tests done in animal tissues, and there is no on-line monitoring of the target location or the magnitude of the temperature elevation. The treatment is therefore sensitive to variations between patients, and the clinical results are variable.

To eliminate these variations, energy delivery and its biological effects should be monitored on line, and exposure should be adjusted to deliver comparable energy to all patients. Researchers from the Brigham and Women's Hospital and Harvard Medical School have worked with engineers and scientists from General Electric Medical Systems to develop an ultrasound surgery system combined with magnetic resonance imaging. This makes on-line temperature information available for monitoring and controlling energy delivery. The status of this project is reviewed in the article.

Ultrasound Induces Thermal Effects

Ultrasound is a mechanical wave with a frequency above the audible range that propagates by the motion of particles so that a pressure wave travels along with the mechanical disturbance. The major advantage of using ultrasound for deep heating is that it penetrates tissues well, at wavelengths on the order of a millimeter. An ultrasound beam can be focused deep into the body onto a spot only a few millimeters in diameter. When the beam is thus focused, ultrasound emitted by a transducer passes through the skin over a wide area at intensities that cause no damage and then converges into a small spot at the focus.

As an ultrasound save propagates through tissues, part of the energy is absorbed and converted to thermal energy. The temperature elevation of the tissue caused by energy absorption is inversely proportional to the beam area. The greatest temperature elevation is induced at the focus, where it can be several hundred times more than in the overlying tissues. This allows tissue at the focal spot to be selectively destroyed while temperature elevation in surrounding tissues is mall, often less than 1°C.

Sharp focusing also allows fast energy delivery, so that temperature levels that cause proteins to coagulate and cells to die can be reached in only a few seconds. The short exposure produces sharp temperature gradients, and the transition distance between coagulated cells and undamaged cells is only a few cells wide.

Ultrasound beams can be focused by using self-focusing radiators, lenses, or reflectors. Transducers with many elements, each driven by properly delayed signals, so-called phased arrays, can also be used to focus the beam. Electrical focusing using phased arrays improves control over the ultrasound beam; for example, it can be used to simultaneously generate multiple focal spots. All current clinical systems use spherically curved transducers for energy delivery.

Several detailed studies have investigated the in vivo effects of ultrasound papameters in animal tissue necrosis. Thermal or caviational effects that occur in tissues during sonication were found to depend on the applied intensity, ultrasound frequency, exposure duration, tissue type, and location. The temperature elevation of the tissue and the duration of exposure determine the extent of tissue damage. Thresholds have been established for many tissue and tumor types, showing that it is possible to kill any tissue with a few seconds exposure if temperatures above 60 to 70°C are reached.

Cavitation is the formation and collapse of gas bubbles in tissue. Its mechanisms and effects are not well understood and appear to depend on tissue type and location. Collapse of the bubbles is associated with thigh temperatures and pressures, which result in various degrees of mechanical damage to the tissue. Hemorrhage and blood vessel damage can also occur when cavitation is present. In addition, the gas bubbles distort the propagation of the ultrasound beam, causing uncertainty in determining the treated tissue volume. The thermal mechanisms are much more predictable and better understood than cavitation and have been used in most ultrasound surgery studies.

Focused ultrasound surgery compares well with established minimally invasive techniques such as lasers and radiofrequency ablation, which require a catheter to be inserted in the tissue. With these systems, energy is delivered close to the catheter and spreads by thermal conduction, so that adequate exposure takes several minutes. As a result, the volume of coagulated tissue is variable and depends on the local blood flow and perfusion of tissue.

Furthermore, because of wider thermal gradients, the transition from coagulated tissue to normal tissue is less sharp with interstitial techniques than with ultrasound. Coagulation of a large volume of tissue with interstitial techniques requires multiple needle insertions. Irregular target volumes close to critical structures are also difficult to treat with the same precision as can be achieved with focused ultrasound surgery.

In contrast, ultrasound is completely noninvasive and utilizes short exposures that result in perfusion-insensitive temperature elevations with sharp gradients. Rapid energy input makes tissue coagulation possible around large blood vessels, which would remove thermal energy by convection during long exposures and thus protect tissues within a few millimeters of the vessel walls The focal spot can be made small, and multiple sonications allow good control over the volume of coagulated tissue, even close to critical structures. Finally, the short ultrasound exposures allow temperature changes to be detected quickly by MRI. The main disadvantage of focused ultrasound is that bone and gas prevent beam propagation, so certain tissues such as lungs are difficult to treat.

MRI Thermometry Localizes Focus and Controls Exposure

Magnetic resonance imaging is the only imaging method that provides adequate information for guiding, monitoring, and controlling therapeutic interventions. The good tissue contrast of MRI can be used for defining the target volume. Ultrasound surgery that can be done while the scanner shows on-line images removes target tissue much more precisely than is possible during open surgery. Equally important is the temperature sensitivity of some MR sequences, allowing localization of the focus and control of the exposure. In addition, tissue changes induced by temperature elevation are often visible in the MR images.

Noninvasive MRI thermometry makes use of the temperature dependence of some physical property whose spatial distribution can be visualized. Three tissue properties have been used for this purpose: spin-lattice decay time (T1), molecular diffusion of water molecules, and the proton resonance frequency of water molecules. In vivo thermometry based on T1 measurements during ultrasound therapy has proved difficult because of changes such as edema and vasodilation induced in tissue by elevated temperatures. However, if exposure times are in the 10 to 20 second range, these slow physiologic changes do not have time to strongly influence tissue properties.

The diffusion coefficient technique that quantifies thermal Brownian movement has been shown to work well in phantoms but is highly sensitive to tissue motion in vivo. Motion sensitivity can be reduced by using echo planar imaging techniques that acquire images quickly, but these require special hardware and reduce the signal to noise ratio.

Good temperature resolution has been obtained using the proton resonance frequency (PRF). Changes in PRF induced by temperature elevation are linearly related to temperature and can be mapped by using changes in phase images. The disadvantage of the frequency sift technique is its insensitivity to temperature changes in fat.

The temperature dependence of T1-weighted and PRF (phase shift) images was tested in rabbit thigh muscle in vivo as a method for monitoring noninvasive ultrasound surgery. Both techniques could detect the temperature elevation, but the phase shift sequence appeared to have a better ration of temperature contrast to noise. Both MR sequences are sensitive enough to localize the temperature elevation and to monitor the normal tissue exposure during therapy.

Although in vivo tissues may have greater variation, the accuracy appears to be adequate for localizing the focus with low-power test exposures, to assure that temperatures between 60 and 100°C are reached during a 10-second therapeutic exposure, and to monitor normal tissue temperature for safety. The upper temperature limit has been set at 100°C to avoid boiling and the resulting formation of gas bubbles that could distort the ultrasound beam. Thus the exposure limits are wide enough to accommodate some degree of uncertainty in the temperature monitoring.

A Prototype System Is Being Tested

The feasibility of coagulating tissue under MRI guidance has been shown experimentally in muscle, kidney, and brain tissue as well as in implanted tumors. In these tissues, the test pulses are visible at power levels that do not cause tissue damage. In addition, the tissue volume at the focus can be coagulated without damaging the overlying tissue. The ability to control tissue damage in this fashion is especially important in the kidney, which is highly perfused but lies beneath a layer of muscle that undergoes much less convective cooling by blood flow.

Because propagation of an ultrasound beam is blocked by air or bone, brain sonications require that a piece of bone be removed to create a window through which the beam can pass.

Besides destroying tissues, focused ultrasound can be used to occlude blood vessels. The ultrasound exposures have not yet been optimized, but coagulation of capillaries and larger arteries has been demonstrated in vivo.

Based on results with experimental system, a prototype ultrasound device for surgery of breast tumors was manufactured by General Electric Medical Systems in collaboration with members of the Department of Radiology at the Brigham and Women's Hospital and GE Corporate Research & Development. The ultrasound fields are generated by a single, focused, air-backed transducer mounted in a standard MRI table. The transducer can be moved in the x, y, and z directions by a hydraulic positioning device within the water bath that acts as a coupling medium. The workstation that controls the transducer is programmed to aim the ultrasound beam at a location defined on an MR image.

During a typical treatment, the target volume is first outlined on a series of MRI scans. A low-energy test pulse is aimed at the center of the target volume by selecting the location with a cursor. The workstation registers the target, aims the beam, generates the test pulse, and transfers the temperature-sensitive image obtained during the sonication. If the region of temperature elevation does not overlap the target volume, a correction is made and a second test pulse is generated to verify the alignment accuracy.

After the test pulses, the complete target volume is sonicated using multiple pulses placed so the coagulated volumes overlap. The dimensions of the coagulated tissue volume for each sonication depend on the duration of the exposure and the applied power.

Clinical testing of the prototype system in the treatment of the breast is now in progress. MR images of tumors that had similar contrast uptake prior to treatment show that treated tumors have no contrast uptake 48 hours later, whereas untreated tumors are not affected. It is too early to draw conclusions about the clinical efficacy or toxicity of the focused ultrasound treatment.

Focused ultrasound could replace some present surgical approaches to benign and malignant tumors. Potential sites should have a soft tissue window that allows passage of the ultrasound beam without encountering air or bone, and the breast is one of the best examples.

Secondary liver cancer is a common problem with a poor prognosis, and liver tumors might be amenable to focused ultrasound surgery. However, the available ultrasound window is restricted by the ribs and by abdominal gas. Many prostate tumors could be reached through the ultrasound window created by a full bladder, or rectal applicators could be developed. Kidney tumors could also be treated, as could deep targets in the brain if a piece of skull is removed.

Blood vessel occlusion is useful for treating arteriovenous malformations and some tumors with an identifiable blood supply. It may also be helpful in the control of abdominal, periotoneal, or pelvic hemorrhage and in the treatment of some trauma victims.

Greater Focal Volume and Improved Temperature Monitoring Are Desirable

In all of the clinical ultrasound surgery studies to date, spherically curved, sharply focused ultrasound transducers have been used. These effectively coagulate small volumes of tissue, but the treatment of large tumors requires multiple sonications. Theoretical and experimental studies have shown that a cooling period is required between exposures to avoid temperature elevation in the tissue between the skin and the target. This makes the treatment of large tumors long and expensive if MRI monitoring is used.

AN effective way to shorten the treatment time is to decrease the number of required sonications. This can be achieved by increasing the acoustic focal volume. Ultrasound phased arrays are the most flexible method, and they allow on line control of the ultrasound beam. Electronic focusing has been used in diagnostic ultrasound for years. Computer simulations have shown that arrays of hundreds of elements may be need when large tumors are to be treated. These arrays cam generate many focal spots simultaneously and tailor the sonication pattern for a given tumor, reducing treatment time and exposure to surrounding tissues.

Phased arrays offer an additional benefit in that the focus can be changed electronically instead of repositioning the transducer mechanically. This advantage is important in the depth direction, where the magnet opening has limited space. Electronic focusing will also be useful in the development of special applicators, such as intracavitary arrays for surgery of the prostate.

There is also room for improved MRI temperature monitoring. It is likely that once the thermal surgery system is used routinely, additional resources will be dedicated to improving the monitoring sequences. A significant improvement would be to obtain a three-dimensional temperature map in the same time as the present 2D maps. A 3D map would allow more precise on-line control to assure accurate target volume coverage and safety of normal tissues. Fast on-line temperature calculation would allow the acoustic energy to be controlled during each pulsed, so that accurate thermal exposures could be delivered without repeated sonications.

Combining focused ultrasound beams with MRI allows precise and reliable destruction of deep tissue volumes noninvasively. The technical feasibility of constructing such a treatment system has been demonstrated, and the technique has shown promise in animal experiments. However, clinical experiments are just beginning.

Treatment protocols have to be developed and tested in patients before wide clinical use will be possible. Clinical trials will provide new information that will be used to improve the equipment so that surgery can be safely and effectively performed. In addition, the clinical experience will establish guidelines for the optimal use of this new technology.

MRI-guided and monitored ultrasound surgery may eventually replace many open procedures by offering increase precision while practically eliminating blood loss and disturbance to normal tissue. This should translate to improved outcomes and significant cost reduction because recover time and hospital stay will be shorter.