JOURNAL OF X-RAY SCIENCE AND TECHNOLOGY 4, 323-333 (1994)

Use of Monochromatic X Rays in Medical Diagnosis and Therapy

What Is It Going to Take? ~

F R A N K C A R R O L L

The Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, Tennessee 3 7232; and the Department of Physics and Astronomy,

Vanderbilt University, Nashville, Tennessee 37235

Received December 22, 1993; revised June 29, 1994

Soon the Vanderbilt University free electron laser will be used to produce near- monochromatic x-ray beams using the phenomenon ofCompton backscattering. This tunable beam has an attractive geometry and flux suitable for use in both diagnostic medical imaging and "chemoradiotherapy."

The earliest diagnostic trials of the efficacy of such radiation beams center on im- proved breast cancer diagnosis, lung tumor discrimination, and dual energy subtraction angiography of the cardiac, pulmonary, carotid, and intracerebral circulations.

Schemata for dual keV x-ray generation, both serial and simultaneous, are under study, along with newer configurations for data acquisition.

Early results have been promising in the study of breast biopsy specimens, but expanded biopsy, animal, and human trials are needed to establish the sensitivity and specificity of the process.

Radiation dose reductions and improved conspicuity of lesions are projected using painless, non-invasive methods.

Compact sources, newer optics, and improved detectors will need to be integrated into any such system destined for installation in a hospital or clinical setting.

In order to stimulate an interchange of ideas between those in the medical field and individuals involved in the design, construction, and operation of newer radiation sources, a monochromatic x-ray workshop was convened, and the content of this article reflects the hopes, dreams, practical concerns, and pleas for congealing multi- disciplinary teams to focus on practical solutions. It is not meant to be a review of the world literature and tends to "beg the issue" in some areas in order to editorialize and stimulate discussion.

With recent technological advances the production of near-monochromatic x-ray beams in configurations of high flux and with attractive geometric proportions has

1 This work was supported by a grant from the Eastman Kodak Corporation, Health Sciences Division, Rochester, NY. The Vanderbilt University Medical Free Electron Laser Center is supported by a grant from the Strategic Defense Initiative Organization/Office of Technology Applications and Office of Naval Research, Contract N00014-87-CO 146 Mod P00003.

323 0895-3996/94 $6.00 Copyright © 1994 by Academic Press, Inc. All rights of reproduction in any form reserved.

324 FRANK CARROLL

become a reality. One would think that potential uses for such beams are limited only by the boundaries of one's own imagination and by the perimeter delimited by the training absorbed in one's given field. However, the mere availability of monochromatic x rays does not assure their usefulness in any field.

One source of monochromatic x rays is the Vanderbilt University's free electron laser (FEL). Using the phenomenon of Compton backscattering, near-monochromatic beams of x rays are produced by colliding the infrared (IR) output of the FEL head- on with the near-relativistic electron beam used to power the FEL. Since the FEL is tunable, the x rays produced are tunable, as well. Hard x rays in the 15-35 keV range can easily be produced, although high fluxes will be available only in the 14-18 keV range initially. This beam has a geometry and flux that make it attractive in both diagnostic and therapeutic areas.

MEDICAL USES

Since Rrentgen's time (1895), the spectrum of a diagnostic x-ray beam has remained little changed, with the exception of beam filtration and better collimation. Softer x rays offer little useful information in the diagnostic arena and contribute extensively to the absorbed patient radiation dose in most examinations. Much harder x rays undergo a greater percent of Compton scattering relative to photoelectric interaction within the tissues, delivering useless, off-axis, softer, somewhat random radiation to the detector, obscuring detail. The greatest promise for medical diagnostic mono- chromatic x rays lies in their tunability throughout the useful diagnostic range, with a narrow bandwidth centered on the frequency of greatest usefulness for the region of the body being imaged.

In the present practice of medical diagnostic imaging, there are areas ripe for sig- nificant improvement, namely those of breast imaging, lung tumor evaluation, and angiography.

Breast

While extremely useful in the early detection of breast cancer, standard low-dose screen-film mammography suffers from some lack of sensitivity and specificity. A small number of cancers will not be evident on the films and many lesions that show clearly are not malignant at all. In some facilities up to 20% of patients with abnormal mammograms must undergo a biopsy procedure to ascertain whether or not the ab- normalities seen represent cancer.

Prior work has shown that breast cancers absorb radiation more avidly than normal breast tissues at low x-ray energies (20-30 keV) (1). More recent studies extend that work to the 14-18 keV range, confirming absorption at higher rates in that portion of the spectrum as well (2).

These studies on linear attenuation were performed on biopsy specimens in vitro. Whether a greater or lesser attenuation occurs in a living, breathing patient remains to be seen. Tumors within the body exhibit certain changes which can enhance ab- sorption, such as an elevated hematocrit (percentage of the blood which consists of red blood cells relative to the overall volume of blood) within the vessels in a tumor. Tumors have a propensity for developing neovascularity, which leaks water into the

M O N O C H R O M A T I C X RAYS IN MEDICAL USES 325

tissues and leaves the red blood cells within the lumen of the vessel, causing a rise in hematocrit in the blood which remains within the vessels. However, leakage of water from the capillaries into the surrounding interstitium or spaces between cells must also be considered. The increase in iron in the concentrated blood would increase linear attenuation, but the increase in water locally would reduce that effect. In the trials on human biopsy specimens, the linear attenuation of the x rays was higher in the more cellular tissues (namely the cancers), but the number of specimens was not high enough to ascertain whether this effect was secondary to a higher effective Z caused by the increase in DNA and RNA in cells with multiple sets of chromosomes (malignant cells) or merely secondary to the greater number of cells relative to other tissue elements between the cancer cells (such as collagen, fat, blood vessels, and ducts). Expanded tests on biopsy specimens, on human tumors implanted in mice (a living system), and eventually on humans in whom biopsies have already been planned due to mammographically suspicious lesions will all be required to determine how valid the alterations in linear attenuation values noted in the laboratory are when applied in the real world of breast cancer screening.

Use of tunable monochromatic beams should allow one to differentiate between cancerous and normal tissues, but only if information is acquired three-dimensionally.

While much of the conventional wisdom in mammography at the present time concentrates on the form of the tumor and either primary or secondary radiographic signs, continuing to rely on the same time-worn methodology perpetuates the lack of sensitivity and specificity in breast diagnosis. If we are to make inroads toward earlier and more accurate diagnosis of breast lesions, we will need to refine the x-ray beam, improve our detectors, improve the manner in which we collect the data and the manner in which we interpret the data when imaging the breast.

Since a standard radiograph is a two-dimensional representation of a three-dimen- sional object, the density of breast tissue (and subsequently its shape) portrayed at any point on a mammogram is a reflection of the additive effects of linear attenuation by each of the tissues traversed by the beam. It is easy to see why tumors may be missed under these circumstances where various tissues are mixed in different proportions (Fig. 1). In the past, linear, circular, and hypocycloidal tomographies of the body were performed by moving an x-ray tube and film in opposite directions about a fulcrum

Patient 1 WN L CA FAT

10x .1 1 0 x . 2 1 0 x . 0 5

_ ; _ , 3.5 CA = Cancer

FAT = Fat

Patient 2 WNL CA WNL

12.5 x .1 5.0 x .2 12.5 x .1

[ ] 1.25 1.0 1.25

3.5

FIG. 1. If an x-ray beam had to traverse 30 cm of tissue in each of two patients, the cumulative linear attenuation effect observed at a given pixel on a film behind the tissues shown would be the same even though both have different sized cancers. The numbers indicate centimeters of the tissue type shown "t imes" the linear attenuation coefficient for each block of tissue. One could imagine similar situations in which a patient without cancer could exhibit the same attenuation as a patient with cancer.

326 FRANK CARROLL

point (Figs. 2A and 2B). Additionally, "infinite tomography" (3) was accomplished using a stationary film changer and a movable tube using variable registration of partially exposed films after their development in a film processor. Similar techniques were also used to develop "digital tomosynthesis" which coupled stationary detectors, digital encoding, and moving x-ray tubes. Key to all of these methods, though, was the use of a moving x-ray tube geometry. However, it would appear that most sources of monochromatic x-ray beams will not be very movable initially. If volumetric data acquisition is desirable (as we feel it is) in discerning attenuation characteristics of various tissues, then innovative approaches will be needed.

Three-dimensional acquisition allows for a layer-by-layer examination of the breast tissues that can even be done on a voxel-by-voxel basis by a properly programmed computer. The machine can search for high attenuation voxels and flag these for closer scrutiny by the radiologist. Such "artificial intelligence" used in conjunction with evaluation by experienced radiologists is being implemented on an experimental basis at Vanderbilt.

If both patient and detector are moved through an arc during exposure by a stationary beam source, the "digital tomosynthesis" method can still be used (4) (Fig. 3A). Im- provements in x-ray optics would also allow "apparent" tube motion by reflection of the x rays to either side of the patient. This could be coupled with motion of the film or use of a stationary digital detector, allowing for the same tomosynthesis techniques to be used (Fig. 3B). These optics are often promised, but have not yet become reality.

Additionally, monochromatic beams can reduce radiation dosage due to the trough in absorbed dose in an optimal range where the photoelectric absorption curve and the Compton absorption curve cross over one another percentage-wise. The absence of very soft x rays that normally are absorbed in the skin and breast tissue will reduce the dose. These also contribute nothing to the imaging process. Higher energy photons which proportionately undergo greater Compton scatter are also absent and therefore

U n m

e / I \1 \ \ / / ! ' f ~ \ .'L? ,'!.. ~ !', ,.'" C B A

u o i

B A FIG. 2. (A) Standard linear tomography utilizes an x-ray tube moving above the patient from point A to

point C. The film below the patient moves in the opposite direction, with film and tube interconnected and pivoted about a fulcrum point. Structures within the patient that are at the level of the fulcrum pivot point will remain in focus at points A, B, and C, while those above or below it will be blurred out when their image is smeared over the length of the film. (B) Digital tomosynthesis uses a moving X-ray tube and stationary image sensor or film that captures multiple discrete images at each tube location and stores each image in computer memory or on a different piece of film. These images can be used to synthesize new images of structures at all levels from front to back using computer algorithms.

MONOCHROMATIC X RAYS IN MEDICAL USES 327

0 s 0 s

A B A B C

Semi-circular D ig i t a l Tomography with Hard Tomosynthesis X-ray Optics

A B

FIG. 3. (A) A stationary monochromatic x-ray source (S) could be used for digital tomosynthesis by moving both the patient (large dark oval) and the detector (at A and B) in a semicircle during exposure. (B) Development of improved "hard" x-ray optics would allow apparent motion of the x-ray source by mirrors first deflecting x rays to the fight (yielding an image at A) and then to the left (image at C), as well as allowing the beam to directly traverse the patient (image B). The film could be moved from A to B and then C, or a stationary detector could be used as in Fig. 2B.

do not fog the x-ray film or confuse the x-ray detector. This latter fact is partially responsible for anticipated improvement in resolution as well.

Lung

When a "spot" (or lung nodule) is detected in a patient's lung on a chest x ray, it is difficult or impossible to determine its etiology. If no previous film can be found with which to compare, the patient usually undergoes computerized tomography (a CAT scan). If the lesion is still suspicious, biopsy is performed, sometimes by surgically opening the chest cavity. All biopsy procedures have morbidity and mortality associated with them, so they are not undertaken lightly.

It stands to reason that lung nodules should absorb monochromatic x-rays differently than normal tissues, in a manner similar to that seen in breast tumors, but it remains to be seen if we can discern significant differences between benign tumors and malignant ones. Work is now underway at Vanderbilt to study such differences in linear atten- uation.

Illumination of suspicious lesions at several different keVs appropriately selected could also allow a "quasi-chemical" analysis of the abnormality and offer the possibility of subtraction imaging. Since normal tissues and tumors absorb similarly at higher keVs and differently in the 14-30 keV range, two exposures (e.g., at 20 keV and 40 keV) could be made and the resultant images subtracted, leaving an image weighted toward the cancerous tissue's image.

It is doubtful, however, that an x-ray spectral analysis, as we now perform them, will yield useful information in vivo due to the "soft" x rays produced and the short path lengths of the photons produced. This is also compounded by the fact that most physiologic atoms or molecules are of low Z and have relatively low energy K-shell

328 FRANK CARROLL

values. However, use of "tagged" molecules, which accumulate in higher concentrations in tumors than in normal tissues, might be considered. Nonradioactive metal "tags" could be used and detected by K-edge bracketing, thereby reducing the whole body dose from radioactivity and reducing the necessity for radiopharmaceuticals and for their handling. Unfortunately, this would require tags with a high tumor-to-background ratio and better detectors than those now available. We would, most likely, be quantum limited, not by the x-ray flux available but by the acceptable radiation dosage that could be given to the patient in our attempts to discern subtle differences created by the presence of the tag.

Vasculature

Dual-energy digital subtraction angiography has been a goal that has been difficult to achieve, due to the polychromatic nature of the standard x-ray beams used. With monochromatic beams tuned above and below the K-edge of iodine, one can take great advantage of the selective absorption of photons at or slightly above the K-edge and use much lower doses of iodinated contrast materials, and possibly visualize the arterial circuit from a venous injection. Radiologists have long been aware of the fact that the arteries can be visualized from venous injections even without the use of dual energy subtraction techniques. In fact, this was done routinely for hypertensive ex- cretory urography to ascertain whether or not renal arterial stenotic lesions were present. A faint, but still discernible, opacification of the aorta and renal arteries was easily accomplished with a rapid intravenous injection. Using temporal, spatial, and energy subtraction, lower-dose studies should be relatively easily achievable.

Such low dose studies could be useful in coronary, pulmonary, carotid, and cerebral circulations.

Dual energy Compton backscattering may be accomplished in several ways, but a simple method might consist of splitting the free electron laser's primary photon output beam. A portion of the beam (e.g., 30%) could be diverted into a passive resonator cavity slightly off-axis from the electron beam path. The remainder of the beam (70%) could be passed through a frequency doubling crystal and thence into a second off- axis cavity (Fig. 4) or it could be directly counterpropagated to the e-beam following the path normally used for single energy production. Alternating keVs could be created by chopping the input beams appropriately (Fig. 5), or produced and collected si- multaneously (Fig. 6). If one desires keVs that are just immediately above and below the K-edge, one need only use a single energy IR input beam split 50-50 to pass half exactly along the e-beam path and the other half slightly off-axis since this latter beam will produce a slightly lower keV x-ray output beam (5, 6).

For arteries and veins to be visualized by x ray requires injection of an intravascular contrast agent that is water-soluble, relatively non-toxic, quickly excreted, and of high enough Z to absorb significant primary radiation from the x-ray beam transilluminating the patient's body. On the whole, less of the contrast material is better. Monochromatic beams can be used to greatest advantage if tuned just above and just below the K- edge of the atom to be injected. One need not expose one high-keV image and one low-keV image for each subtraction image to be generated if the region imaged is reasonably stable and unmoving over a period of seconds. Significant radiation dose

MONOCHROMATIC X RAYS IN MEDICAL USES 329

beam-spUtter / Freq. Doubling Crystal Infrared~=._~ • ~ • ~,.or Beam " .I.'% 1"%

T mirror E-B..=amA mirror ¥ mirror r ~ i ~ ~ mirror

Etalon 1 ~ . i ~ . Interaction Zone

E-beam.~...''~ X-rays dump •

FIG. 4. Two different keVs may be produced by the Compton backseattering process by several means. This figure depicts only one of these. One could produce two energies (one at half the keV of the other). If, for example, 30% of the infrared beam were injected into a passive resonator cavity (etalon 1), flux would build up within the cavity and x rays would be generated at the interaction zone that were dependent upon the frequency of the light and the energy of the e-beam. If the other 70% of the infrared beam were passed through a frequency doubling crystal and thence into a second etalon, x rays would be generated at twice the energy of the first. Both keVs could be produced simultaneously or in rapid succession by appropriate beam choppers.

savings can be garnered from the individual subtraction of multiple high energy images from one low-keV picture.

The K-edges of iodine, gadolinium, technetium, barium, silver, zinc, copper, and iron all lie within easy reach of our tunable monochromatic source. Many FDA- approved drugs and contrast agents already exist which incorporate these atoms, so that the development of new drugs is not necessary. The medical community, ra- diologists, and pharmaceutical companies know a great deal about these drugs and they can fit readily and immediately into these monochromatic trials without years of initial safety testing.

17 keY

J\ keY

I 34 k i

ke~

One keV at a time

. . . . . . . . . . . . . . . . . . i i i i i i ' - ' . i i i i i i i i i i

minus ~ -" ~ Image at 17 keV Image at 34 keV Subtraction

FIG. 5. If each monochromatic beam were used individually and serially to acquire an image of a patient with an x-ray contrast agent in the vessels, it is possible to subsequently subtract the image frames by computer, leaving the difference images that take advantage of K-edge effects.

330 FRANK CARROLL

17keY 34keV

I J ~ J~ Dual keV Simultaneously I Exposed and ACluired keV

m i n u s - -

°ptiX'ray ~ "' 'S'ubi'r'acii'o'n"

i i i . . . . . . . . . . . . l l | 34 keV FIG. 6. Both keVs can be produced at the same time and simultaneously acquired by two different detectors

whose images would be spatially separated by their energy after reflection from an appropriate x-ray optic. Since the perspective of the x-ray optic would be stationary to any patient or object placed between the stationary x-ray beam and itself, the separate energy images could be subtracted without parallax errors to yield the final dual-energy digital subtraction image (such as that obtained in an arteriogram).

THERAPEUTIC USES

In a manner analogous to the selective interactions observed when visible and IR lasers are focused on solid surfaces, within bulk materials, and now in soft tissues, monochromatic x rays also hold promise for such selectivity.

The coupling of drugs which seek out tumors and the selective activation of these agents by monochromatic beams tuned to the heavy metals custom fabricated within the drug molecules promises "more bang for the buck" in a new form of chemora- diotherapy (7-10).

Porphyrins which accumulate in and around tumors, monoclonal antibodies which seek out specific receptors on cell surfaces, and deoxyuridines which are rapidly taken up in the DNA of rapidly dividing cell nuclei can all be fabricated to contain metals whose k-shell electrons can be targeted by tunable monochromatic x rays. Ejection of these electrons can cause ionization regional to the compound but confined to a very short distance (e.g., 7-10 urn). Depending upon the drug used, the ionization damage can destroy vessels supplying the tumor, disrupt the cell membrane, or lethally damage the intranuclear DNA needed for cell replication (Figs. 7A-7C). While ion pairs will be produced in surrounding tissues as well, these events will be more evenly distributed throughout the tissues than the more concentrated effects obtainable by tuning to the specific frequency that will interact with the k-shells of multiple drug molecules ad- sorbed to the specific cellular target.

MACHINE RELATED

Generation of monochromatic x rays sometimes seems the easiest factor to address in the equation for coupling production ofx rays and bringing them into an environ- ment such as a hospital, where they can safely be used on patients. If energetic electrons are used to produce the x rays and their axis of flight is in line of sight to the patient,

MONOCHROMATIC X RAYS IN MEDICAL USES 331

Molecules

Hematoporphyrins Monoclonal Antibodies

A B C

FIG. 7. Custom-made drugs with an affinity for specific sites in and around tumor cells can be activated by externally administered monochromatic x rays releasing k-shell electrons locally for enhanced lethality regional to the cancer. (A) Drugs may target the fine new feeding vasculature around a tumor, (B) adhere to specific antigenic sites on the cell surface, or (C) incorporate into the DNA and RNA in the cell nucleus.

they must be controlled and diverted away from the patient, but this is risky should control systems fail. Alternatively, x rays produced can be reflected away from the e- beam axis (11, 12), but large reflectance angles become problematic beyond 10 keV. The promise of "hard" x-ray optics seems to always be with us, but has not yet become reproducible reality. Kumakhov optics, new multilayer technologies, and other usable grazing incidence optics show theoretical promise but have been "underwhelming" in the practical arena. A hybrid system using both e-beam deflection and smaller x- ray reflectance angles would seem the most likely to succeed in a commercial machine. With smaller x-ray reflectance angles, greater flux can be delivered to the patient examination area, allowing for slightly less stringent demands on the accelerator used in the system.

In the atmosphere of cost containment, we cannot afford to graft another expensive technology onto the already overburdened health care system unless we expect it to replace some older technology or we can prove that its benefit-to-cost ratio is so great in curing the ills of man that it would be far more costly not to opt for adopting it.

It is of little value to society as a whole to develop expensive technologies for the diagnosis or treatment of diseases afflicting very few individuals. However, technological developments that will allow the earlier detection of any of the major killers of humans would seem laudable. Monochromatic x rays theoretically should be able to significantly improve screening for breast cancer, lung cancer, heart attacks, and strokes. Even small increments in improvement in these areas take on monumenta l proportions in the overall scheme of health care delivery and preventive medicine or improved therapy.

The earliest devices available for monochromat ic x-ray generation are frequently multiuse machines that are somewhat large, expensive, and inaccessible. Their progeny, however, can be designed and built for compactness, lower cost, single-use, and "port- ability,"with the capability for installation in a hospital setting.

332 FRANK CARROLL

Likewise, it is not enough to simply "cost out" the price of building a machine to produce these monochromatic beams and accept that as the bottom line. We must be more realistic in factoring in the ancillary costs attached to their implementation. Site preparation, yearly maintenance costs, consumption of expendable supplies, sal- aries of trained personnel, loan amortization, insurance, upgrades, and useful life are only a few of the factors which can be far more costly to an institution than the "sticker price" even of a relatively inexpensive machine. These costs must be balanced against the lean years of a trial period during which the efficacy of the technology is proven to the skeptical and sometimes cynical patients, physicians, and third-party insurance carriers. Reimbursement for equipment costs and study interpretation is unlikely during the first few years of such clinical trials.

Newer x-ray optics will also be required to ensure flexibility and safety and to further contribute to compact design.

Detectors with near unit efficiency at x-ray photon energies will be required to keep the radiation dose to a minimum, while accumulating data at high resolution and at rapid frame rates.

Monochromatized industrial x-ray tubes, it would seem, lack the broad tunability, flexibility, and pulse structure obtainable with electron beam sources.

Rapidly switching the kVP (kilovolts potential applied to an x-ray tube) with electron bombardment of dual metal anodes makes for cumbersome, expensive single-use ma- chines, again without broad tunability and the micro- and macropulse structure of a machine such as the FEL.

ANCILLARY CONSIDERATIONS

Production of monochromatic x rays will open the door to other important tech- nologies as well. Medicine will, no doubt, significantly benefit from x-ray microlith- ography and its resultant effects on further miniaturization of computers and remote sensors. We may also be blessed with improved nondestructive testing of joint prostheses for the body (such as hips, knees, and shoulders), detecting internal micro- cracks which would have led to failure of the new joint after implantation.

From the data available to date, it would seem that no single method of x-ray generation would be universally useful due to the needs of "industries" as diverse as microelectronics and medicine.

CONCLUSION

If we can clearly define both diagnostic and therapeutic scenarios that will benefit from monochromatic x rays, then it behooves us to draw together the significant talents of theorists and designers in machine development, optics, and detectors, in order to pool their energies and those of the users to optimize carefully thought-out systems. Joint workshops, multidisciplinary interaction on a daily basis, and pooling of ideas and data will benefit not only the scientists involved, but all of us as patients, and in the long run humanity as a whole.

REFERENCES

1. P. C. JOHNS AND M. J. YAFFE, Phys. Med. Biol. 32, 675 (1987). 2. F. E. CARROLL, J. W. WATERS, W. W. ANDREWS, R. R. PRICE, D. R. PICKENS, R. WILLCOTT, P. C.

TOMPKINS, C. ROOS, D. PAGE, G. REED, A. UEDA, R. BAIN, P. WANG, AND M. BASSINGER, Invest. Radiol. 29(3), 266 (1994).

MONOCHROMATIC X RAYS IN MEDICAL USES 333

3. E. R. MILLER, E. M. MCCURRY, AND B. HRUSKA, Radiology 98, 249 (Feb. 1971).

4. D. R. PICK~NS, Personal communication, Dept. of Radiology and Radiological Sciences, Vanderbilt University School of Medicine.

5. C-M TANG, B. HAFIZI, AND S. K. RIDE, "Thomson Backscattered X-Rays from an Intense Laser Beam," Naval Res. Lab. Rep. NRL/MR/6791-93-7328 (July 26, 1993).

6. F. E. CARROLL, J. W. WATERS, R. R. PRICE, C. A. BRAU, C. F. Roos, N. H. TOLK, D. R. PICKENS, AND W. H. STEPHENS, ll'Ivest. RadioL 25(5), 465 (May 1990).

7. S. J. ADELSTEIN, Am. J. Roentgenol. 160, 707 (1993).

8. U. P. SCHMIED, J. m. NELSON, F. L. STARR, AND R. SCHMIDT, Invest. Radiol. 27, 536 (1992).

9. R. J. LANZAFAME, D. W. ROGERS, J. O. NAIM, H. R. HERRARA, AND J. R. HINSHAW, Lasers Surg. Med. 7, 280 (1987).

10. H. J. VREMAN, B. C. EKSTRAND, AND D. K. STEVENSON, Pediat. Res. 33, 195 (1993).

11. W. D. ANDREWS, E. E. CARROLL, J. W. WATERS, C. A. BRAU, R. R. PRICE, D. R. PICKENS, P. A. TOMPKINS, AND C. F. ROOS, NucL lnstrum. Methods Phys. Res. A 318, 189 (1992).

12. P. A. TOMPI~NS, C. A. BRAU, W. DONG, J. W. WATERS, F. E. CARROLL, D. R. PICKENS, AND R. R. PRICE, Proc. IEEE, in press.