Search

The sections in this article are 1 Introduction 2 Concepts for Spatial Localization 3 Image Intensity Dependence on Imaging Parameters 4 Image Intensity Dependence on Signal 5 Imaging Techniques 6 Image Contrast 7 Resolution, S/N, and Imaging Time 8 Conclusion 9 Acknowledgements 10 Biographical Sketch Related Articles

The sections in this article are 1 Introduction 2 Conceptual Limits on Field Strength 3 Magnet Selection 4 Imaging Results 5 Conclusions 6 Biographical Sketch

Purpose To study the effects of T1, T2 changes due to field difference on imaging contrast C and signal difference-to-noise K between a certain tumor model, mammary adenocarcinoma, and muscle. The results may provide a guide in optimizing the imaging sequences for MR imaging at different magnetic fields.

Magnetic resonance (MR) imaging at 0.063 T, 0.17 T, 0.35 T, 2.94 T, and 4.85 T was performed in rats with implanted mammary adenocarcinomas (and in control rats) to study hydrogen density and the effects of magnetic field strength on T1 and T2 relaxation times in vivo. T1 increased by an order of magnitude and T2 decreased by a factor of two between the lowest and highest field strengths. Only about half of the increased water content of tumor, compared with that of muscle, was reflected in increased hydrogen density. The sensitivity of relaxation time change was increased at the lower field strengths. These results are of significance in understanding the effects of field strength on lesion detectability, contrast, and signal-to-noise ratio.

Magnetic resonance imaging (MRI) offers a unique and extraordinary window into the body. Since 1983 the growth and acceptance of this modality as a diagnostic tool has been prodigious, and the enthusiasm with which it is received still grows. The original uses were centered on head imaging, and have since expanded to spine and extremities. Some acceptance is now being seen for body imaging, more for central portions of the chest and in the pelvis than in the abdomen. The routine use of MRI for cardiovascular work is generally limited to large centers. Because of the patient demand for MRI, and because of the desire to obtain revenue from the units, MRI has not been fully exploited in areas where the primary payoff is a contribution towards understanding disease processes. For this latter purpose, one of the most important characteristics of MRI is that it is non-invasive, and therefore ideally suited for both cross-sectional and longitudinal studies. There is an obvious match between MRI and the needs of those who aim to understand the evolution of the human atherosclerotic plaque.

At least one extra NMR measurement cycle is performed without any imposed magnetic gradients during readout and recordation of the NMR RF response. Calibration data derived from this extra measurement cycle or cycles can be used for resetting the RF transmitter frequency and/or for phase shifting other conventionally acquired NMR RF response data to compensate for spurious changes in magnetic fields experienced during the NMR data measuring processes. Some such spurious fields may be due to drifting of the nominally static magnetic field. Another source of spurious fields are due to remnant eddy currents induced in surrounding conductive structures by magnetic gradient pulses employed prior to the occurrence of the NMR RF response signal. Special procedures can be employed to permit the compensation data itself to be substantially unaffected by relatively static inhomogeneities in the magnetic field and/or by differences in NMR spectra of fat and water types of nuclei in imaged volumes containing both.

Nuclear magnetic resonance (NMR) images of live rats with sterile and pyogenic abscesses, hematomas, and various implanted and spontaneous neoplasms demonstrated good contrast differentiation between pathologic and surrounding normal tissues. This differentiation was maximal when both the T1 and T2 tissue relaxation times were used as criteria. Neoplasms have a broad range of T1 and T2 values and may be confused with abscesses or hematomas. Tissue rate constants (1/T1 and 1/T2) are mainly dependent on total water content, the exception being fat, which has a 1/T2 value much shorter than that expected on the basis of water content alone.

Advances in imaging technique have improved the efficiency of clinical nuclear magnetic resonance (NMR) imaging, and will allow total patient examination time that equals or is more favorable than that of x-ray computed tomography (CT). The whole head can be examined with NMR in a 6.5-minute imaging time with a spatial resolution of 1.7 mm. Fifteen sections in the body can be similarly imaged. Quantitative T2 ("spin-spin" relaxation time) information, as well as estimates of T1 ("spin-lattice" relaxation time) can be obtained in this time. Quantitative T1 information requires an additional procedure.

Two dilute oral iron solutions, made from commonly available nonprescription dietary supplements, were found to enhance the gastrointestinal tract in nuclear magnetic resonance imaging of live rats and one human volunteer. The paramagnetic and pharmacologic properties of ferric ammonium citrate were more favorable than those of ferrous sulfate heptahydrate. The paramagnetic iron solutions shorten T1 and T2 relaxation times of water protons in the contrast media-filled gastrointestinal tract, producing easily observable change in NMR intensity. Because these iron solutions are available commercially and are known to be well tolerated, the clinical use of iron-containing NMR contrast agents for the gastrointestinal tract is feasible.

Practical constraints make it difficult to build large-aperture echo-planar magnetic resonance (MR) imagers. The implementation of a pediatric imager and its performance are described. Spatial resolution and signal-to-noise levels comparable to those of 1982 state-of-the-art MR imagers have been achieved in imaging times of 0.05-0.15 seconds. T1 and T2 information are obtainable in the echo-planar mode. A major issue is that of chemical-shift displacements.

Thirty healthy subjects and 15 patients with a variety of musculoskeletal disorders were examined by conventional radiography, computed tomography (CT), and nuclear magnetic resonance (NMR). NMR proved capable of demonstrating important anatomic structures in the region of the lumbosacral spine. Lumbar disk protrusion was demonstrated in three patients with CT evidence of the disease. NMR appeared to differentiate annulus fibrosus from nucleus pulposus in intervertebral disk material. Avascular necrosis of the femoral head was demonstrated in two patients. The cruciate ligaments of the knee were well defined by NMR. Muscles, tendons and ligaments, and blood vessels could be reliably differentiated, and the excellent soft-tissue contrast of NMR proved useful in the evaluation of bony and soft-tissue tumors. NMR holds promise in the evaluation of musculoskeletal disorders.

Advances in technology make it possible to extend the range of physical and chemical tissue parameters used for formation of images. This in turn adds to the capabilities to diagnose disease and assess its extent. Quantitative improvements in instrumentation augment the fidelity with which the imaging is performed, and at a certain point provide a quantum jump in diagnostic capability. We discuss recent technological advances in digital radiography, NMR and nuclear medicine imaging that demonstrate these effects.

Encoding the Precession phase angle of proton nuclei for Fourier analysis has produced accurate measurement of fluid velocity vector components by MRI. A Pair of identical gradient pulses separated in time by exactly ½ TE, are used to linearly encode the phase of flow velocity vector components without changing the phase of stationary nuclei, Two-dimensional Fourier transformation of signals gave velocity density images of laminar flow in angled tubes which were in agreement with the laws of vector addition. These Velocity profile images provide a quantitative method for the investigation of fluid dynamics and hemodynamics.© 1985 Academic Press,Inc.

Computational techniques allow the simulation of NMR images as well as the construction of tools for evaluating the impact of changing parameters. Use of these tools assumes the acquisition of a minimal set of patient data from which the tissue properties T1, T2 and hydrogen density H, can be taken. Changes in technique, imaging parameters such as TE, TR or TI, and field can be simulated and examined individually as well as concurrently for their effects on the contrast between a lesion and its surrounding tissue. This evaluation is done by plotting the absolute difference in signal intensities as a function of the imaging parameters of the technique. Retrospective calculations of intensity images at new values of the pulse timing parameters may show improved diagnostic utility.

The efficiency and quality of three-dimensional Fourier transform (3DFT) magnetic resonance imaging were evaluated and compared with those of 2DFT imaging. For a fixed imaging time and number of sections, 3DFT imaging with conventional spin echoes always had a worse signal-to-noise ratio (S/N) than did 2DFT imaging, when both were optimized with respect to choice of repetition time (TR). With partial-flip magnetic resonance (MR) imaging, S/N was nearly equal for 2DFT and 3DFT imaging when both were optimized with respect to TR and flip angle. 3DFT imaging can have cross-section artifacts that exceed those of 2DFT imaging. For very thin sections these artifacts may be lessened, and 3DFT imaging can achieve this with smaller gradient pulses. Over-all, 3DFT imaging was found to be advantageous only for the very-thin-section imaging and in combination with partial-flip MR imaging.

Although cross-sectional magnetic resonance examination of the head and body is useful for screening large regions of tissue, subsectional regions of the head and body often need to be examined. Orthogonally directed, selectively irradiated planes with different flip angles produce a spatially limited signal region from which two- or three-dimensional volume images can be reconstructed. Images with limited fields-of-view can be acquired in reduced imaging time. We present a general description of this technique. These subsectional or "inner volume" images eliminate respiratory motion artifacts by excluding moving tissues from the imaged volume. A result of this technique is a high signal from rapid pulsatile blood flow, produced without cardiac gating the pulse sequence.

NMR imaging of rats with implanted hepatomas in the liver demonstrates that under the imaging conditions of this study tumors of over 8 mm2 in area can be detected with high accuracy. Compared to normal liver, approximately 70% of these tumors had a combination of relaxation times (T1-T2) that could be uniquely identified as tumors, while the other 30% demonstrated relaxation time combinations that overlapped those previously found for abscesses, brain, and hematomas.

To determine whether hydrogen magnetic resonance imaging at 0.7 T provides added clinical value over imaging at 0.35 T, images of the heads of patients with various intracranial disorders were obtained at these field strengths. Measurements of tissue contrast (C), signal-to-noise (S/N) ratio, and T1 and T2 relaxation times were determined. For a given spin-echo sequence with equal imaging time, resolution, and data sampling window, the product C X S/N was somewhat lower for the lower field strength. Under conditions of imaging with equal chemical shift artifact, C X S/N at 0.35 T was equal to or greater than that measured at 0.7 T. With an increase in field strength, T1 of pathologic areas and surrounding normal tissues increased, resulting in a corresponding loss of absolute signal level and decrease in contrast. Lesions were equally well seen at both 0.35 T and 0.7 T. The increased T1 and decreased C X S/N for higher magnetic fields--when measured with a fixed imaging time, resolution, chemical shift, and sequence--suggest that such field strengths may not improve tissue contrast, diagnostic ability, or clinical throughput when compared with lower field strength systems.

Theoretical analysis predicts that performing magnetic resonance (MR) imaging with partial (less than 90 degrees) flip angles can reduce imaging times two- to fourfold when lesions with elevated T1 values are being examined. This time savings occurs because repetition time (TR) is reduced when imaging is performed with partial flips. Partial flip MR imaging can also improve signal-to-noise ratio (S/N) in fast body imaging. For this study, analytical tools were used to predict image contrast and S/N for short TR, partial flip sequences. Experimental implementation of the short TR, partial flip sequences that analytical work had predicted would be optimal supported the analytical predictions and demonstrated their validity. Partial flip MR imaging is applicable to reducing imaging time only when the ratio of signal differences to noise exceeds threshold values in conventional MR images. Partial flip sequences can be used to advantage in MR imaging of both the head and the body, and the observed effects are predictable through theoretical analysis.

A study of 24 rats implanted in the hind leg with mammary adenocarcinomas and five sham-implanted rats, followed from the second to the eight week postimplantation, showed nuclear magnetic resonance imaging capable of detecting all the tumors without yielding any false-positives in the control rats. The T1 relaxation time of tumors overlapped that of muscle, and the T2 times overlapped fat, but the combination was unique when comparing tumors to muscle and fat. Necrotic regions of the tumor and the bladder contents tended to have very long T1 and T2 relaxation times. The difference in relaxation time between tumors and muscle could be accounted for in terms of water content, which was approximately 8% higher for the tumors. The study corroborates data from previous studies indicating that NMR imaging is a highly sensitive modality, although T1 and T2 times are not exclusive indicators of malignancy.

Fibrosarcoma tumor cells were implanted in the hind legs of 25 rats and studied from the first to the sixth week postimplantation. Nuclear magnetic resonance imaging detected all tumors and did not yield any false-positives in five control rats. The T1 relaxation values of tumors overlapped those of muscle, and the T2 values overlapped those of fat, but the combination of the two values allowed discrimination of each of the three tissues with no overlap. The difference in relaxation time between tumor and muscle could be accounted for on the basis of water content, which was approximately 14% higher in the tumors. This study confirms data from previous studies suggesting that nuclear magnetic resonance imaging is a highly sensitive modality, but that T1 and T2 values are not specific for individual pathologic conditions.

Human images of excellent quality have been obtained with Nuclear Magnetic Resonance (NMR), using two different techniques: projection reconstruction (PR) and two-dimensional Fourier transformations (2DFT). It has been shown in CT that the reconstruction algorithms will multiply the noise in the projections and change the noise spectrum. A comparison of the noise of the two techniques is presented under the constraint of equal imaging conditions using both computer simulations and data obtained with the UCSF human imager. The computer simulations show that the two techniques produce approximately equal signal to noise values when S/N ~ 15. Below that value small differences are observed. For both simulations and actual data, the 2DFT shows a uniform spatial frequency response while the PR method shows a roll-off at low spatial frequencies. The 2DFT method is shown to be more resistant to artifact formation.

The appearance of intraluminal signal in the cardiac chambers, the descending aorta, and blood vessels was studied in healthy subjects and patients with myocardial disease on first and second spin-echo gated magnetic resonance images. Signal was present in the cardiac chambers and the aorta at various phases of the cardiac cycle when physiological or pathological slow flow conditions are expected in healthy subjects and in patients. Healthy individuals tended to show signal in the ventricles and aorta during end-diastole, and signal was less likely to be present at higher heart rates and in systolic images. In patients with regional or global left ventricular dysfunction, intraventricular signal tended to persist into systole. Surprisingly, intraventricular signal was not present with increased frequency adjacent to infarcted regions of the myocardial wall. Thus, the mere presence of intracavitary signal cannot be used as an indicator of either regional or global cardiac contraction abnormalities. In the left atrium, signal was often present during systole. Physical factors determining the appearance of signal of flowing blood are discussed in an Appendix.

Ten patients with definite multiple sclerosis underwent hydrogen nuclear magnetic resonance imaging with a 3.5 kilogauss superconducting magnet, using the inversion recovery and spin-echo techniques of signal acquisition. Results were compared with high-resolution x-ray computed tomography. Spin-echo images demonstrated abnormal regions as areas of variably increased signal intensity. The contrast between abnormal and normal white matter improved as the intervals between sequential radiofrequency pulses and between pulse administration and signal sampling were increased. Inversion recovery images demonstrated abnormal areas as regions of decreased signal intensity but did not visualize lesions as well as spin-echo imaging. Spin-echo and inversion recovery imaging each demonstrated more extensive abnormalities than did computed tomography.

Eleven Sprague-Dawley rats with experimentally produced infarction of the lower extremities were imaged by nuclear magnetic resonance (NMR) immediately after infarction, and 6 and 24 hours later. Contrast between the infarcted and control muscles was noted immediately and continued to increase through 24 hours. These changes corresponded to a significant increase in the water content of the infarcted extremity at 24 hours. These results suggest promise for NMR imaging in the identification of myocardial infarction and should encourage further investigation in this area.

This article provides some examples of what the previous article outlined. The sections on Image Display and Myocardial Characterization contain numerous illustrations of gating, cross-sectional images, flow signals, and ischemic injury. The possibility of metabolic imaging with NMR and quantitating blood flow is also considered.

The principles underlying nuclear magnetic resonance (NMR) are explained, and its use in image formation is described. The discussion also covers: multislice imaging; image resolution and contrast; effects of flow on images; hardware for magnetic resonance imaging; magnets; magnetic gradients; RF coil, transmitter, and receiver; computer system; and spectroscopy.

A technique is described for obtaining tomographic images of hydrogen distribution in animals using nuclear magnetic resonance (NMR). Resonant frequency is proportional to magnetic field strength, so that spatial resolution is achieved by frequency selection and magnetic field shaping. The results of scanning a phantom and two rats are presented.

The normal urinary bladder and several benign entities of the bladder were examined in 50 patients by magnetic resonance (MR) imaging. Specific features assessed included appearance of the bladder wall, optimal repetition (TR) and echo delay (TE) parameters for bladder-wall demonstration, and differentiation among various benign abnormalities, including bladder-wall hypertrophy, inflammation, and mucosal congestion, on MR images. The bladder wall in the 30 healthy subjects was best displayed using a TR = 2 sec, TE = 56 msec image, which gave 60% contrast between the bladder wall and urine and 48% contrast between the bladder wall and fat. Demonstration of bladder-wall hypertrophy required similar imaging; bladder distention was necessary to demonstrate the thickness of the bladder wall. Congestion and inflammation were best demonstrated on TR = 2 sec, TE = 56 msec images, which gave 45% contrast. Normal and/or hypertrophic bladder wall were distinguished from inflammation and congestion on the basis of signal intensity variations and/or T1 and T2 relaxation parameters.

Gated magnetic resonance (MR) imaging was used to evaluate central cardiovascular anatomy in 172 subjects, 31 of whom were healthy volunteers. Using the spin-echo technique, images of diagnostic quality were obtained in 93% of cases with TE = 28 msec and in 65% of cases with TE = 56 msec. Transverse multisection sequences encompassing most of the left ventricle required approximately 6-8 minutes. Corroborative studies were available in 134 of 141 patients who had cardiovascular disease; two dimensional echocardiograms and angiography in 133 and 100 patients, respectively. Gated MR demonstrated the wall thinning and complications caused by prior myocardial infarctions and high signal intensity of the myocardium at the site of acute myocardial infarctions. MR accurately demonstrated anatomic abnormalities owing to hypertrophic and congestive cardiomyopathies, congenital abnormalities of the heart and great vessels, rheumatic heart disease, pulmonary hypertension, and cardiac and paracardiac masses. Depiction of cardiovascular anatomy and pathoanatomy was attained without the use of any contrast media. Consequently, gated MR is an effective technique for cardiac diagnosis. The short time required for tomographic examination of the entire heart using the multisection technique renders this a practical cardiac imaging modality.

A new type of electronically driven viscometer with digital readout is described. The advantages of this instrument are that only small samples are required for analysis, fluid perturbations are minimal, cyclical measurements can be carried out rapidly and the system is small and moderate in cost.

This retrospective study of 144 patients was made to (a) assess the potential of magnetic resonance (MR) for demonstrating lymph nodes using spin-echo technique, (b) compare the MR results with those of CT, and (c) determine the optimal pulse-sequence interval (TR) and echo-delay time (TE) for imaging lymph nodes. The reported CT findings on normal lymph nodes were compared with MR findings in 60 patients who underwent MR imaging of the neck (20 patients), chest (20 patients), abdomen (10 patients), and pelvis (10 patients) for conditions other than lymph node disease. The results showed that CT is presently better than MR for imaging neck and abdominal lymph nodes less than 13 mm in diameter. The ability of MR to demonstrate normal-size (less than 10 mm) lymph nodes in the chest was comparable to that of CT. In addition, MR scans of 84 patients with proven abnormal lymph nodes (8 neck, 49 chest, and 27 abdomen and pelvis) were assessed: in 72 patients, these nodes had also been imaged by CT. MR and CT gave similar results with abnormal lymph nodes (greater than 13 or 15 mm), but MR displayed these nodes better because of its excellent soft-tissue contrast resolution. MR can clearly differentiate abnormal lymph nodes from normal fat, muscle, vessels, adult thymus, thyroid, and diaphragmatic crura, as well as from primary tumor and lymphoceles. Optimal demonstration of lymph nodes with MR required two sequences: one with a short TR and one with a long TR and long TE. Preliminary results indicate that MR holds great promise for the demonstration of lymph nodes in every part of the body.

Thin-section magnetic resonance imaging at low field strengths requires analysis of the relative merits of data collection techniques for direct three-dimensional versus multisection two-dimensional imaging. This analysis was done with specific emphasis on shortened relaxation times and the use of reduced magnetic field gradients. Three-dimensional Fourier transform techniques can provide thin-section images of good diagnostic quality when combined with partial flip-angle gradient-reversal techniques.

This paper represents a preliminary study of the effects of regional temperature distribution in two-dimensional nuclear magnetic resonance (NMR) T1 imaging. It is found, as expected, that variations in local temperature appear as variations in the corresponding T1 image. The potential use of NMR T1 imaging in temperature measurements is evaluated in the case of water and blood samples. Using containers where the temperature could be either known or directly controlled with reasonable accuracy, images are obtained with samples having at least two regions at different temperatures. As expected, T1 is found to vary linearly with 1/T over the range of 0 degrees C to about 40 degrees C for blood. The potential use of T1 imaging in hyperthermia applications is also discussed.

To test if changing magnetic fields and radiofrequency fields used in nuclear magnetic resonance imagers could induce electrical currents capable of causing localized tissue heating in metal surgical clips and prostheses, steel surgical clips, copper wire clips, and hip prostheses were exposed to fields greater than those used in the nuclear magnetic resonance imager. Observations indicated that no significant heating should be expected from implanted surgical clips during exposure. The heating of larger metallic implants should be further investigated.

The sequence of radiofrequency and magnetic field gradients used to form an NMR image produces changes in the signals emitted by moving blood when compared to the signals from stationary blood. These changes can be used to measure the open area of vessels. In addition, specific relaxation time signatures of lesions may help in identifying the nature of the obstructions.

In the very short time since magnetic resonance imaging (MRI) was born it has gained surprisingly rapid and enthusiastic acceptance and has speedily proliferated, particularly in the United States and Western Europe. Magnetic resonance imaging (MRI) has successfully challenged computed tomography (CT) in all areas of the body where respiratory motion does not degrade the image (Steinberg, 1986). Newer techniques using a multiplicity of approaches are starting to close the gap between CT and MRI, even in the upper abdomen where the effects of respiratory motion are most pronounced. Although MR is already widely clinically applied and is an accepted everyday diagnostic modality in most large medical centres in the United States, it is not a mature modality. It is rapidly evolving, with whole new areas opening to investigation which will vastly broaden its applications.

The multiparametric nature of NMR imaging allows the same section to be displayed in many different ways by either adjusting acquisition parameters or calculating additional images, creating a possibly overwhelming data set. The need for this many images may be reduced by creating tissue type maps where each tissue is identified according to its characteristic NMR parameters and painted using a color code. Although T1 and T2 values for normal tissue are reproducible, variability among patients, the wide range of T1,T2 and N(H) values for abnormal tissue and partial volume averaging make the determination of tissue signatures difficult without significant operator intervention. A hierarchical processing algorithm, pyramidal segmentation, allowing simultaneous analysis at multiple levels of spatial resolution provides the basis for a tissue mapping algorithm with significantly reduced operator supervision.

Whole body nuclear magnetic resonance (NMR) imaging of the cardiovascular system was carried out in early clinical trials in 244 volunteers and patients using a 3.5 KGauss (0.35 T) unit. The spin echo technique with multiple imaging parameters was used. Blood vessels were clearly discriminated from solid organs and lesions because little or no intraluminal signal is seen with laminar blood flow at normal velocities, whereas a more intense image is generated by solid organs. Characteristic flow signals were observed in normal patients and were accentuated by varying the imaging parameters. Cardiac chambers were well delineated in some patients on nongated images. In one case, internal topography of the ventricles was exquisitely displayed on a gated image. Intraluminal pathology, such as dissection of the aorta, aneurysms of the aorta and left ventricle, and aortic atheroma, was clearly demonstrated. Patency of coronary arterial bypass grafts was shown. Abnormal flow patterns due to slow or turbulent flow were accentuated on images using the second spin echo. This preliminary experience indicates the considerable potential of NMR imaging in the evaluation of cardiovascular diseases.

A programme was written to detect flow in a sequence of MR images and to construct 3-D vessel maps from the detected flow regions. Reduction in first echo intensity and prolongation of calculated T2 value were used as flow-detection criteria. The programme was used to image the aorta in 6 patients with abdominal aortic aneurysm and 5 patients with aortic dissection. Results were compared with contrast angiography in 9 cases and X-ray CT in 2. The images obtained by the two techniques were comparable in 9 cases. In 2 cases of aortic dissection, where flow in the false lumen was slow, the false lumen could not be demonstrated by angiography, but was clearly seen in the reconstructed MR images.

The potential clinical importance of T1 and T2 relaxation times in distinguishing normal and pathologic tissue with magnetic resonance (MR) is discussed and clinical examples of cerebral abnormalities are given. T1 and T2 values may be used in three ways: (a) Relative values, obtained by an analysis of intensity images with varying dependence on T1 and T2, may be used if absolute values for T1 and T2 are not required for diagnosis. (b) If an absolute value is desired, the numerical values for the relaxation times may be generated using a region of interest on the intensity images. (c) In cases in which both T1 and T2 change may require a calculated image to indicate the contribution of each to the signal intensity, the numerical value may be used to generate analogue images of T1 or T2 calculations. Five patients with cerebral infarction, 15 with multiple sclerosis, two with Wilson disease, and four with tumors were imaged. Hemorrhagic and ischemic cerebrovascular accidents were distinguished using the spin echo technique. In the patients with multiple sclerosis, lesions had prolonged T1 and T2 times, but the definition of plaque was limited by spatial resolution. No abnormalities in signal intensity were seen in the patient with Wilson disease who was no longer severely disabled; abnormal increased signal intensity in the basal ganglia was found in the second patient with Wilson disease. Four tumors produced abnormal T1 and T2 relaxation times but these values alone were not sufficient for tumor characterization.

All instruments used to produce nuclear magnetic resonance (NMR) images have a number of common features. A strong, uniform, and stable magnetic field has to be imposed through the subject. This field has to be varied in space and time to define the location of pixels in the image. A radio frequency magnetic field is applied to excite the nuclei. The radiofrequency signal from the nuclei is detected and its strength determines the intensities of the pixels in the image. The imaging process, data acquisition, and display are all controlled by computers. Within this general structure many options exist for collecting the data necessary for image reconstruction. Areas of recent improvement include increased numbers of slices in a given time, finer spatial resolution and sensitivity to blood flow.