Real-time interactions in MRI.

With the advent of magnetic resonance fluoroscopic imaging, it is possible to obtain images at a rate which can be considered effectively real-time. Image refresh rates can be as low as 0.5 s. Such rates are fast enough to allow MR to be used for guidance of surgical instruments during micro-surgical procedures. Examples include the localization of biopsy needles within a tumor, localization of surgical tools for application of therapeutic drugs or the proper positioning of optical fibers for application of laser heat. It may also be used for long duration monitoring of a patient in an attempt to diagnose rare and short duration clinical events.

Analysis of hybrid imaging techniques.

Hybrid imaging techniques have been proposed as a means of decreasing imaging time without the cost and technical constraints of echo planar imaging. With this technique phase encoding measurements acquired at different echo times in more than one multiecho experiment are used to form a single image. This study analyzes the trade-offs which occur in the selection of this technique over more conventional imaging in terms of signal/noise, contrast, resolution, imaging time, and efficiency. Hybrid imaging is shown to be advantageous when raw speed is essential such as in abdominal or pediatric imaging. When coverage rather than time becomes the important factor, hybrid imaging does not offer a significant advantage over conventional methods. T2 decay will also serve as a roll-off filter which will reduce the spatial resolution, but not the noise, for short TE hybrid imaging.

Fluoroscopic MR imaging at 0.064 Tesla.

The authors developed a system for ultrafast magnetic resonance imaging (MRI) protocols at low field. The system design permits the acquisition of the raw data in the background while the reconstruction and display steps repeat as fast as they can in the foreground. The performance speeds depends partly on the desired use. By collecting raw data at a rate of 20 ms per echo with an echo delay of 9 ms, a complete data cycle for a 128x64 image takes 1.28 s. However, once half of that data is incorporated into the reconstruction, the image appears complete. Using this set of parameters the authors were able to get the rate of the recon/display loop to paint about two times per completed raw data cycle, showing an entirely new image at least once per second with an apparent frame rate of two per second. Interleaving of two or three orthogonal scans reduces the speed of update but provides better information. The authors discuss the system design for rapid scan/recon/display and demonstrate the image quality available at low field strength with scan times below one second.

Measuring signal-to-noise ratios in MR imaging.

The signal-to-noise ratio (S/N) in magnetic resonance imagining is one of the variables that must be measured when comparing the relative performance of different techniques. Although various investigators and official groups have proposed different methods for measuring S/N, these are generally not practical for use by a physician working in a clinical situation. The authors present a simple method that should serve for estimating S/N in most cases.

Directions in magnetic resonance imaging technology.

Over the past few years there have been substantial improvements in the performance of magnetic resonance (MR) imagers. As image quality improved it became possible to perform studies in less time, increasing the throughput and the availability of the technique. A crucial contributor has been improvements in signal/noise. Techniques can now give signal/noise levels that a few years ago would have required much longer imaging times. With partial flip angle imaging techniques, it is possible to maintain image contrast and signal/noise while using reduced values of TR which decrease imaging time. It is also possible to decrease the number of acquired data lines and replace these lines mathematically at reconstruction time. Signal/noise is sacrificed, but the benefit is almost a factor of two in acquisition time. Echo planar techniques provide even higher speed imaging. In addition to the trade-off of signal/noise versus acquisition time, signal/noise can be traded for reduction in magnetic field strength. This results in reduced cost, improved patient access and also offers reduction in motion artifacts.

Characterization of MR spectroscopic imaging of the human head and limb at 2.0 T.

Using section-select and phase-encoding gradients, the authors obtained phosphorus chemical shift images of the human head and limb. Phosphorus spectra were acquired from planar sections divided into voxels as small as 7 cm3 in calf muscle and 27 cm3 in brain, with total examination times, including setup and proton locator imaging, of roughly 1 hour. Both spin-echo and free induction decay (FID) methods were employed; the FID gave superior results. Signal-to-noise ratios for the beta-adenosine triphosphate and phosphocreatine resonances were as high as 10:1 and 13:1 from volumes of 27 cm3 in brain.

Bayesean-deblurred Planar and SPECT nuclear bone imaging for the demonstration of facial anatomy and craniomandibular disorders.

Ambiguities in diagnoses can often be resolved when images from different imaging modalities are compared, and when images are processed with algorithms that improve resolution and contrast. Bayesean deblurring algorithms were developed and applied to Planar and SPECT images of the maxillofacial and temporomandibular joint regions. The combined use of Planar and SPECT imaging with Bayesean deblurring were complementary and provided more diagnostic information than either modality individually. A facial imaging protocol using Planar and SPECT imaging and Bayesean deblurring is described. SPECT maxillofacial anatomy is presented, as well as the application of the imaging protocol of craniomandibular dysfunction. Although not recommended for all patients with craniomandibular disorders, combined use of Planar and SPECT images and Bayesean deblurring techniques appears to be useful in diagnostically difficult or refractory cases.

Emission imaging of patients with craniomandibular dysfunction.

Signs and symptoms of craniomandibular dysfunction in 37 patients were compared with the results of corrected cephalometric tomography and an emission imaging protocol consisting of both planar and single photon emission computed tomography (SPECT) (7500 ZLC Orbiter) images. The planar images and the single photon emission computed tomography projection views were processed with a bayesian deblurring algorithm to improve image quality. The correlation of emission imaging with craniomandibular dysfunction, as indicated by temporomandibular joint pain and joint noise, showed a high sensitivity (93%) and a high specificity (86%), whereas the correlation of corrected cephalometric tomography with temporomandibular joint pain and joint noise showed a relatively high sensitivity (89%) but a low specificity (27%). These results indicate that emission imaging is a sensitive and accurate indicator of craniomandibular dysfunction.

Signal-to-noise ratio and section thickness in two-dimensional versus three-dimensional Fourier transform MR imaging.

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.

Magnetic resonance imaging of the TMJ disc in asymptomatic volunteers.

Forty-two temporomandibular joints (TMJs) in 21 asymptomatic volunteers were visualized by magnetic resonance imaging (MRI). The subjects, 12 males and nine females, were between 23 and 43 years of age and had no history of TMJ pain, joint noise, limited opening, or previous treatment for TMJ disorder. A cephalometric head-holder was designed to position the TMJ in an accurate and reproducible manner and multisection parasagittal images were obtained perpendicular to the longitudinal axis of the condyle. MR images depicted anterior disc position in 32% of the asymptomatic joints (8/24 males, 5/18 females). Anterior disc position in asymptomatic subjects may be a predisposing factor to TMJ dysfunction or simply an anatomic variant whose prevalence must be considered when evaluating TMJ dysfunction.

The utility of principal component analysis for the image display of brain lesions. A preliminary, comparative study.

Principal component analysis (PCA), a common tool from multivariate statistical analysis, has been implemented into the computer display system of a MR imaging device. PCA allows the calculation of images in which the information in a defined region of interest inherent in the basic acquired images is condensed. PCA image calculation has been applied to acquired MR studies of 13 patients with brain lesions. The appearance of the brain lesions on the resultant PCA images was scored in comparison to the acquired images before and after administration of Gd-DTPA as well as to other calculated images including T1, T2, hydrogen density, and contrast-optimized images. The conspicuity of a lesion and the number of distinguishable components within a lesion were slightly superior on PCA than on the acquired images. PCA is an analytical tool for MR imaging that should be helpful in revealing information that is inherent in, but not readily visible on, standard acquired MR images.

The value of relaxation times and density measurements in clinical MRI.

The hope that MRI relaxation time signatures would identify tissues, specifically, malignancies, has not been realized. This is due much less to measurement inaccuracies than to a large intrinsic variability and overlaps between malignancies and many benign pathologies. Neither has there been success in predicting relaxation times from basic tissue compositions. Nevertheless, MRI provides a qualitative measure of tissue hydration, and of flow, on the basis of relaxation times. Furthermore, pixel-by-pixel maps of relaxation times have proven useful in understanding the MRI process, in predicting the efficacy of untried techniques, and replace, in many circumstances, the need for acquisition of images with diverse sequencing parameters.

Partial flip angle MR imaging.

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.

MRI of blood flow: correlation of image appearance with spin-echo phase shift and signal intensity.

Phase-sensitive imaging was used to correlate signal distribution with phase shift and velocity distribution in spin-echo magnetic resonance imaging (MRI). Flow-dependent, changing intensity patterns that were seen in a constant-flow phantom study were explained by the simultaneous effects of inflow signal enhancement, first-echo dephasing, and outflow signal loss occurring during laminar flow. In clinical studies, first-echo dephasing was shown during laminar flow in the inferior vena cava. Turbulent flow was demonstrated in the descending thoracic aorta during late systolic flow, and turbulent dephasing-rephasing was shown in the abdominal aorta.

Information processing in magnetic resonance imaging.

Techniques of image processing have been developed for the extraction of information from magnetic resonance images. The response of nuclei to a sequence of magnetic stimulations has been modeled and used to predict the effects of changing magnetic resonance parameters on signal intensity. This allows the calculation of new images from a small set of acquired data without additional acquisition. These images can be used to predict the results of new imaging techniques before actual implementation and to simulate the effects of parameter variations which are unfeasible or impractical, such as variable field strength. New forms of representation, for example, a tissue type map in which each homogeneous tissue category has been identified and labeled, can be used for more direct interpretation.

Variable magnetic resonance imaging parameters: effect on detection and characterization of lesions.

With change in the imaging technique and magnetic field strength used in magnetic resonance imaging, wide variations in the delineation of pathologic features occur. Using imaging data from patients with known pathologic conditions, we evaluated the intensity images in spin-echo and inversion-recovery imaging at varying repetition times, echo times, and inversion times over broad ranges and changing magnetic field strengths. Differences in conspicuity and the apparent size of the lesions are important to consider in diagnosing and evaluating pathologic conditions, especially when different imagers and techniques are employed.