In This Issue
The Vertiginous March of Technology
March 1, 2000 Volume 30 Issue 1

Functional Biomedical Imaging

Wednesday, December 3, 2008

Author: Thomas F. Budinger

    New engineering will enable diagnosis and advance treatment of the major health problems.

This article focuses on the engineering facets that are key to the development of the human imaging technologies of X-ray computed tomography (X-ray CT), magnetic resonance imaging (MRI), positron emission tomography (PET), and single photon emission computed tomography (SPECT). These technologies were selected to demonstrate future plausible developments through the use of examples of imaging brain function, heart function, arthritis, vertebral disc disease, prostate cancer, breast cancer, and colorectal tumors. Ultrasound and other imaging methods not included herein are discussed in Mathematics and Physics of Emerging Biomedical Imaging, a publication by a National Research Council panel on mathematics and physics in emerging biomedical imaging techniques (Budinger et al., 1996a).

PET gives images of the distribution of an injected positron-emitting radionuclide usually attached to a compound such as an analogue of sugar. The radiation dose is usually less than the dose that individuals receive from natural background each year. These radionuclides have short half-lives (e.g., 2 to 109 minutes) and are produced in cyclotrons or reactors. When the positron emitter (e.g., carbon-11, oxygen-15, nitrogen-13, or fluorine-18) decays, the positron (a positive electron) interacts with a local electron and the masses of the positron and electron are converted to two high-energy photons that travel in opposite directions and are detected by a scintillation crystal and a photoelectron multiplier tube. The detector material and electronics are key elements for the present and future development of this technique.

At present, the detector is a scintillator that converts the high-energy photon into light, which is converted to an electric current by the photoelectron multiplier tube. Small arrays of detectors with newly invented scintillators sandwiched between a photodiode array and an array of photoelectron multiplier tubes have now been developed that can achieve spatial resolution of 2 mm for brain imaging. It has taken 22 years to reduce the dark current from 200 nanoamperes to 20 picoamperes for low-noise photodiode arrays (Budinger et al., 1996b). In the future, avalanche detector systems may replace photoelectron multiplier tubes, allowing construction of very compact systems.

The major attribute of PET and SPECT that distinguishes them from MRI and X-ray CT methods is the high sensitivity with which they can detect metabolic activity and trace the concentration of specific proteins in the body (e.g., neuroreceptor proteins of the brain) (Budinger, 1995; Cherry and Phelps, 1995). PET scans detect tracers in the nanomolar concentration range, allowing studies of diseases such as schizophrenia, manic depressive diseases, and Alzheimer’s disease.

An example of the image of glucose metabolism reflected by the accumulation of fluoride-18 deoxyglucose, an analogue of glucose, is shown in Figure 1 (see full text pdf for figures and tables). In the brains of people with Alzheimer’s disease, there is a decrease in metabolism in the parietal regions of the cerebral cortex. This is one of the regions of the brain that processes memory. Understanding the neural chemistry that leads to this decrease in glucose metabolism requires a detector system with much higher resolution than exists today, because the areas that control this region are deep in the brain and in the range of 2 mm in size.

Another example of the specificity of the PET technique is the definition of the dopamine system, which is the neural component of the brain affected by Parkinson’s disease. Resolution improved from 17 mm in the mid-1970s to 2.6 mm in the mid-1990s (Table 1), and further improvements can be expected in the near future. Current levels of resolution will enable studies of abnormal brain chemistry in diseases such as manic depressive disorders, schizophrenia, and other mental disorders.

Magnetic Resonance Imaging

The three categories of magnetic resonance now being used are magnetic resonance proton imaging, magnetic resonance spectroscopy (MRS), and functional MRI (fMRI) (Chen and Ugurbil, 1999). The magnetic resonance image is an image of the distribution of hydrogen nuclei associated with water (Figure 2). The contrast between gray and white matter in these images is due to the rate of signal decay, which is dependent on the local tissue environment. MRS has the unique capability to detect the chemical composition of tissues, but the method requires concentrations of substances to be in the millimolar range. By contrast, emission techniques (e.g., PET) can detect tracers a million times less concentrated than general magnetic resonance. New engineering leading to higher-resolution fields will result in improved MRS sensitivity. In 1986, Paul Lauterbur and the author started the effort to create a national resource whole-body magnet at 10 teslas (T) that would allow studies of compounds having less than 1 millimolar concentrations. The need for high-resolution fields for brain studies are discussed below.

Functional Brain Imaging

Three types of dynamic brain function are quantifiable by noninvasive imaging: blood flow changes in response to sensory or mental activity, neurochemical activity, and metabolic activity of energy consumption (e.g., glucose and oxygen consumption). Radiotracer techniques have shown that blood flow in the capillary bed of small regions of the brain increases 2 to 30 percent when that region is called on to increase nerve activity by some external or internal stimulus. The more recent advent of MRI methods for detecting a signal change associated with this local activity makes it possible to measure mental functioning noninvasively without the use of radiation. Thus, there has been a surge of activity in the area of fMRI.

PET and fMRI were compared for their ability to detect activation of the area of the brain important for processing noxious images or odors (Irwin et al., 1996; Zald and Pardo, 1997). Immediately after the stimulus arrived in the appropriate cerebral cortex area, PET showed flow or volume by the rate of accumulation of a flow tracer. MRI showed a signal that reflects the change in the local magnetic environment because of a difference in the amounts of oxy- and deoxyhemoglobin in that area. Oxyhemoglobin is diamagnetic and deoxyhemoglobin is paramagnetic. The concentrations of oxyhemoglobin and deoxyhemoglobin in local regions change, and these changes are visualized by subtracting the image after activation from the image before activation. Activation of neurons is followed by an increase in neuron metabolism.

The results of the fMRI method are expressed in terms of the so-called BOLD (blood oxygen level dependent) effect. A momentary decline in oxyhemoglobin of 0.5 percent negative BOLD effect was seen recently at 4 T; blood flow then increased to compensate for the loss of local oxygen. The positive BOLD signal is the result of an overflow, and the signal seen in most studies at 1.5 T is not from the precise area activated, but is rather from the arterial and venous blood draining the activated region. This functional activation can be confusing to the neuroscientist who wants to know exactly which nerve cells are being activated, and the need of better signal-to-noise ratio has led investigators to use stronger and stronger magnetic fields. Using 4-T rather than 1.5-T fields clearly showed that there is an initial drop in signal followed by a reactive flooding of the area with blood, resulting in more oxyhemoglobin being delivered than is needed to sustain or keep ahead of the neuronal metabolic demands.

From a clinical standpoint, fMRI has applications in identifying areas of the brain to be avoided in surgery (e.g., for tumor excision). The scientific future of the technology depends in good part on an improved signal-to-noise ratio. The noise artifacts from heart, respiratory, and patient motion accompanying the activation procedure and other effects are amplified by the after-minus-before subtraction method. This amplification, along with a low signal-to-noise ratio, often results in a need for repeated studies. The electrical engineering methods of signal processing have been successful in retrieving the true signal, but even at 4 T the signal change is only 3 percent, whereas flow changes of 30 percent are quantitatively measured using radioactive methods. Thus, to achieve reliable scientific data when exploring brain function, fields of 4 T and higher are required. Two recent installations at 7 T (University of Minnesota) and 8 T (Ohio State University) will allow studies of brain function beyond sensory and motor function (e.g., mental activity).

The evolution of high-field human MRI from 1.5 T to 8 T or above will enable detection of metabolites by using carbon-13 patterns associated with specific compounds and detection of electrolytes such as potassium and sodium. It is even possible to evaluate the proportion of intra- and extracellular concentration of these ions with quadripolar moments by using multiple quantum methods associated with imaging (Lack et al., 1995; Schepkin et al., 1996; Star-Lack et al., 1997). These methods will allow studies of the mechanisms of diseases such as manic depressive disorders.

Neurochemistry of the brain can also be studied noninvasively by using radioactive techniques such as imaging ligands which depict the activity of the dopaminergic neurons whose abnormal functioning is known in Parkinson’s disease and is suspected in schizophrenia and other mental diseases.

Deaths from coronary atherosclerosis have declined dramatically in the past 20 years. An estimated 1 million lives per year are saved due to a lower incidence of certain risk factors (e.g., smoking and high-fat diet) and by medical interventions (e.g., lipoprotein-modifying medication, antihypertensive drugs, beta blockers, and surgery). Nevertheless, coronary artery disease remains a major cause of early death and a major health-cost factor.

The contemporary methods of evaluating coronary arteries involve X-ray imaging of the coronary arteries after injection of contrast material via catheterization. Catheterization is expensive and has some medical risks and thus is not used as a routine screening method for the detection of atherosclerosis. The early and inexpensive diagnosis of coronary artery disease through advanced computed tomography methods and cardiac MRI is possible in the near future, but only through engineering innovations.

X-ray CT technology recently achieved an important role in detecting coronary artery disease by a noninvasive technique that is able to demonstrate calcium deposits in diseased coronary artery walls. Using electron-beam technology where the beam instead of the gantry is moved back and forth over the cathode allows one to collect data rapidly to avoid the motion artifact of the beating heart. This new type of study is rapid, inexpensive, and uses a low dose of radiation, and the results show pathology in the vessel wall rather than changes in lumen diameter, which can be less sensitive in 50 percent of patients with atherosclerosis but little or no constriction of the coronary vasculature.

MRI Heart Imaging

By using MRI, it is possible to show the flow channels in the heart without catheterization. Three recent developments show the near-term potential of this imaging technique. First, the existence of flowing blood allows the MRI procedure to emphasize differences between moving protons in blood and surrounding stationary tissue; thus, though the use of specialized pulse sequences, the major branches of the coronary arteries can be visualized. This visualization can be enhanced by using magnetic resonance contrast material such as natural albumin with an attached paramagnetic element (gadolinium), which is now in clinical trials. This material is injected intravenously and will stay in the lumen of vessels for a period sufficient to allow visualization of the vessels in the body by MRI methods. The material is now under review by the Food and Drug Administration.

The second enabling technology is the development of a very fast scanning method wherein the magnetic field gradients are switched very rapidly in order to gather image data that can be reconstituted to avoid motion distortion. The third enabling engineering technology is the development of MRI units that allow ease of access for both patient and attendants. Thus, in the near future, we can expect to have pictures of our coronary trees as part of our medical evaluation just as today we have records of the waveforms of electrical activity of the heart, such as the electrocardiogram, in our medical records.

Radionuclide methods of measuring blood flow to the heart muscle have played an essential role in the diagnosis of coronary artery disease. The isotope delivery to the muscle can be a more sensitive measure of flow than anatomy images. However, although diagnoses and modern medical therapies applied before or during the acute cardiac attack have reduced the death rate from coronary artery disease by 22 percent in the past 20 years, the incidence of hospital visits from patients with congestive heart failure (i.e., mechanical output deficit relative to the blood volume flow needed for everyday activities) has gone up by a factor of 4 in the same period of time for patients over 60. There are 400,000 new congestive heart failure cases every year. After the onset of this disease in women, the life span is 3 years; in men, it is 1.5 years. The prevalence is over 4 million. There is no cure for this problem except transplantation or artificial assist devices.

Treating Congestive Heart Failure

Congestive heart failure is caused not only by the stress on the heart muscle due to atherosclerosis, but by untreated hypertension, metabolic or viral diseases, and possibly responses to excessive mental stress. There is a progressive failure of the pumping ability of the heart that has undergone the stress of having part of the muscle sack rendered ineffective by a previous heart attack, or by the results of misrepair (remodeling errors) in response to chronic stress. As the good muscle pumps blood out through the aorta, it also puts pressure on the portion of the heart wall that is unable to contract as well as the good muscle. The stress at the interface between the good muscle and the weak muscle could lead to the same consequences as continued stress on the entire heart muscle from long-term untreated hypertension.

What might be going on in the heart muscle complex? The National Heart, Lung, and Blood Institute recognizes that congestive heart failure is now a major problem in health care, a problem that is very poorly understood and inadequately treated. An engineering approach to this disease promises to reveal both the mechanisms involved and the proper treatment. Although 70 percent of the heart muscle volume consists of muscle cells (myocytes), these cells are only 30 percent of the total cells in the heart. The remaining 70 percent of cells are associated with the blood vessels and the scaffolding that organizes myocytes, which must contract and relax in synchrony and in well-defined directions for efficient pumping.

Collagen, a protein that forms fibers with the tensile strength of Kevlar, is the major material of the scaffolding. Thus, the maintenance of the human heart during normal and stressful functioning operations must involve collagen maintenance. Surprisingly, the turnover time for collagen in the heart is one month, or 3 percent per day. Imagine the situation: constant remodeling of a bridge structure every month while the myocytes contract in synchrony. Congestive heart failure appears to be a problem of collagen scaffold remodeling, but this has not been conclusively demonstrated. After tissue stress (e.g., a twisted ankle or skin scratch) there is an invasion at the site of injury by fibroblasts, which endeavor to repair the injury. Very frequently, this repair effort causes further injury if not controlled by anti-inflammatory agents (Luster, 1998).

New developments in both PET and MRI can play a major role in the understanding and treatment of congestive heart failure. PET can measure the activity of collagen formation by evaluating collagen synthesis or breakdown and the pool sizes of constituent amino acids (e.g., proline) through the use of radiolabeled tracers. MRI methods of monitoring the moment-to-moment changes of segments of the heart wall use a saturating radio frequency pattern such that a null-signal grid appears on the MRI image set. The motion of this grid reflects the twisting, contraction, and dilation of individual sectors of the muscle. This motion allows quantitation of the spatial distribution of the strain tensor and gives quantitative evidence of the progression of the congestive heart failure as well as measures the efficacy of treatments. Thus, the engineering of radioactive molecules to be probes of collagen remodeling, the perfection of emission tomography methods of imaging the beating heart, and MRI imaging of the strain tensor allow an approach to understanding one of the major health care problems facing the nation.

Contemporary methods for detecting and staging human cancers rely heavily on X-ray imaging, particularly for cancers of the lungs, bones, and gastrointestinal tract, where the imaging techniques are used as an adjunct to the symptom-related evaluation of patients. Conventional projection X-ray imaging can be diagnostic in some cancers, but usually more specialized procedures using X-ray CT or X-rays with fluoroscopy during infusion of contrast material, such as barium enema, are used. The new methods of PET, ultrasound imaging, and MRI have brought major advances to cancer detection.

Breast cancer is an example of a problem that begs for new engineering solutions. Digital mammography (Feig and Yaffe, 1995) has not made many expected clinical advances despite the electrical engineering applied to this problem. This is because the intrinsic contrast between diseased tissue and normal tissue is only 1 percent or less. On the basis of suspicious mammograms there are 700,000 biopsies of breast tissue annually, but in only 35 percent of cases is cancer actually present. Thus, some other method is needed to select patients for biopsy.

A new technology for studying breast cancer involves use of cancer-specific radiopharmaceuticals. For example, by using the radionuclide fluorine-18 attached to deoxyglucose and a PET instrument, it is possible to scan the whole body for evidence of cancer and metastases distant from the cancer and to evaluate the effectiveness of therapies. Another new engineering technology being applied to breast images is the use of solid-state components to create an inexpensive compact gamma camera that promises to significantly aid early breast cancer diagnosis (Gruber et al., 1998).

Arthritis

In one way or another, arthritis and osteoporosis affect 13 million people in the United States. Specialized applications of the major imaging methods are being focused on these problems. MRI can detect cartilage and bone matrix changes and has provided a cost-effective substitute for X-ray methods such as myelography. PET can determine the role of the immune system and potentially the metabolism of collagen, which is the scaffold on which bone is formed. X-ray CT can now provide almost real-time three-dimensional images of joints and defects in the neck and back vertebrae.

It appears that stresses in the intervertebral discs result in a disruption of the collagen supporting sheath. These stresses are accompanied by changes of the material within discs that controls electrolyte concentrations important for proper osmotic pressures. The need to understand vertebral disease early in life can be fulfilled by engineering MRI magnets that allow dynamic imaging under load-bearing conditions as well as access to patients for image-guided surgery.

Visualization and Virtual Reality

Image multiplexing (or fusion), image-guided therapy, and virtual reality in human imaging are three current activities in signal processing and visualization that are important to medical science. Straightforward methods of image processing have allowed the superposition of multimodality imaging methods (e.g., MRI and PET) in a presentation of fused images. Although there is no evidence that fusion images have led to major advances in clinical diagnoses, these methods have enabled the image-guided therapy that is playing a prominent role in medical practice. The major engineering innovations involve construction of a new type of MRI magnet that allows open access to the patient during surgical manipulations, such as tumor biopsy. This approach, pioneered by General Electric Corporation, promises to have a significant impact on the efficacious practice of medicine.

One of the most exciting innovations in biomedical imaging is the application of virtual reality to the day-to-day clinical study of human disease. Many years ago, the movie Fantastic Voyage presented the concept of visualizing the body from the inside, traversing blood vessels as though they were freeways. The ability to visualize the anatomy and function of the body by noninvasively entering a three-dimensional image data set and exploring it from the inside out would realize the clinician’s dream of dissecting the living body without knife, needle, endoscope, or injury. This dream is now being realized by virtual colonoscopy in day-to-day clinical practice, and potentially will be available for bronchoscopy and even spinal canal and vertebral foramen exploration.

Through innovations such as spiral CT (Kalender et al., 1990) and MRI, three-dimensional data can be collected in only a few minutes without patient discomfort. Modern, low-cost computers can manipulate three-dimensional data sets using newly developed software. This technology of visualization can dramatically change the practice of medicine as it enables new approaches to understanding cancer of lungs and colon, aortic diseases, spinal stenosis, and lower-back pathologies, which are expected to increase in prevalence in the aging population of the United States.

New developments will continue in the future of biomedical instrumentation for medical imaging. These developments are stimulated by the desire to visualize and quantitate functions of the human body and are enabled by advances spanning electrical, mechanical, and chemical engineering, and the computer and materials sciences. The professional engineers and scientists who will create the future are from these disciplines and the integration of these disciplines is a primary example of the evolving field of bioengineering.


Acknowledgments

The research summarized here is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health and the Office of Biology and Environmental Research of the Department of Energy. Dr. Kathleen M. Brennan assisted significantly in the preparation of this manuscript and the accompanying figures.

References

Budinger, T. F. 1995. Single photon emission computed tomography. Pp. 121-138 in Diagnostic Nuclear Medicine, M. P. Sandler, R. E. Coleman, F. J. T. Wackers, J. A. Patton, A. Gottschalk, and P. B. Hoffer, eds. Baltimore: Williams & Wilkins.

Budinger, T. F., F. Wehrli, S. M. Blumenfeld, F. A. Grunbaum, R. M. Henkelman, P. C. Lauterbur, W. Loeffler, F. Natterer, S. J. Nelson, L. Shepp, R. G. Shulman, and B. M. W. Tsui, eds. 1996a. Mathematics and Physics of Emerging Biomedical Imaging. National Research Council. Washington, D.C.: National Academy Press.

Budinger, T. F., K. M. Brennan, W. W. Moses, and S. E. Derenzo. 1996b. Advances in positron tomography for oncology. Nuclear Medicine and Biology 23(6):659-667.

Chen, W., and K. Ugurbil. 1999. High spatial resolution functional magnetic resonance imaging at very-high-magnetic field. Topics in Magnetic Resonance Imaging 10(1):63-78.

Cherry, S. R., and M. E. Phelps. 1995. Positron emission tomography: Methods and instrumentation. Pp. 139-159 in Diagnostic Nuclear Medicine, M. P. Sandler, R. E. Coleman, F. J. T. Wackers, J. A. Patton, A. Gottschalk, and P. B. Hoffer, eds. Baltimore: Williams & Wilkins.

Feig, S. A., and M. J. Yaffe. 1995. Digital mammography, computer-aided diagnosis, and telemammography. Radiologic Clinics of North America 33:1205-1230.

Gruber, G. J., W. W. Moses, S. E. Derenzo, N. W. Wang, E. Beuville, and M. H. Ho. 1998. A discrete scintillation camera module using silicon photodiode readout of CsI(T1) crystals for breast cancer imaging. IEEE Transactions on Nuclear Science NS-45:1063-1068.

Irwin, W., R. J. Davidson, M. J. Lowe, B. J. Mock, J. A. Sorenson, and P. A. Turski. 1996. Human amygdala activation detected with echo-planar functional magnetic resonance imaging. Neuroreport 7:1765-1769.

Kalender, W. A., W. Seissler, E. Klotz, and P. Vock. 1990. Spiral volumetric CT with single-breath-hold technique, continuous transport, and continuous scanner rotation. Radiology 176:181-183.

Lack, J. M., V. D. Schepkin, and T. F. Budinger. 1995. In vivo four dimensional 1H and 23Na imaging using continuously oscillating gradients. Proceedings of the Society for Magnetic Resonance 326 (abstract).

Luster A. D. 1998. Chemokines--chemotactic cytokines that mediate inflammation. New England Journal of Medicine 338:436-445.

Schepkin, V. D., I. O. Choy, and T. F. Budinger. 1996. Sodium alterations in isolated rat heart during cardioplegic arrest. Journal of Applied Physiology 81:2696-2702.

Star-Lack, J. M., M. S. Roos, S. T. Wong, V. D. Schepkin, and T. F. Budinger. 1997. Four-dimensional 1H and 23Na imaging using continuously oscillating gradients. Journal of Magnetic Resonance 124:420-438.

Zald, D. H., and J. V. Pardo. 1997. Emotion, olfaction, and the human amygdala: Amygdala activation during aversive olfactory stimulation. Proceedings of the National Academy of Sciences 94:4119-4124.

 

About the Author:Thomas F. Budinger, NAE, is head of the Center for Functional Imaging, E.O. Lawrence Berkeley National Laboratory, and professor and chair of bioengineering, University of California, Berkeley. This paper is adapted from a presentation he made at the 1997 NAE Annual Meeting Technical Symposium.