Quality assurance methods and phantoms for magnetic resonance imaging: report of AAPM nuclear magnetic resonance Task Group No. 1

MRI scanner variability studies using a semi-automated analysis system

Due to the unique design of the Parallel Rod Test Object (PRoTO) and the associated semi-automated analysis program, it was necessary to test it extensively for precision and accuracy, and preliminarily for utility, before its distribution for wider use in MRI system quality control (QC). The test object and analysis program measured the desired quantities reproducibly and they accurately measured predicted changes from intentionally adjusted imaging system parameters, yielding sensitivity of the various test measures to deviation in the system operating parameters. From a single scan of the most recent revision of the test object, multiple quantitative quality control measures were obtained throughout the scanning volume on two MR imaging systems over periods of six and twelve months, respectively. From these and earlier trials, an initial indication was obtained of which performance measures are worth monitoring for QC. This experience suggests that signal-to-noise ratio (SNR) and distortion (including display scale) should be monitored but not necessarily the resolution. The latter was only found to alter at the same time or later than other parameters such as SNR had changed. Slice thickness was found to vary on some units and this measure was also used in normalizing the SNR by voxel volume. SNR, distortion, and resolution measurements using field-echo sequences were less stable than those using spin-echo sequences. Use of this QC program to test a wide variety of image quality measures allowed timely assessment of the long-term variability of the units tested. Long-term variability may become among the most important measures for comparison of system performance and maintenance.(ABSTRACT TRUNCATED AT 250 WORDS)

Signal-to-noise, resolution, and bias function analysis of asymmetric sampling with zero-padded magnitude FT reconstruction

This report describes NMR image effects due to sampling asymmetry when using zero-padded magnitude FT reconstruction. With this method, the MTF is not flat over the spatial frequency passband, so resolution cannot be accurately described by a single variable such as voxel size. At small to moderate asymmetry, shortened (reduced window duration) asymmetry provides increased S/N and decreased resolution, whereas shifted (constant window duration) asymmetry yields essentially constant S/N with simultaneously increased and decreased resolution. A bias function expression describes image distortion due to sampling in terms separable from the imaged object. The analyses are consistent with previous descriptions of perceived image differences related to data asymmetry.

A comparison between magnetic resonance imaging and computed tomography for stereotactic coordinate determination

The spatial accuracy of magnetic resonance imaging (MRI) has not been established for stereotactic surgery. Magnetic susceptibility artifacts may lead to anatomical distortion and inaccurate stereotactic MRI coordinates, especially when targets are in regions of the brain out of the center of the magnetic field. MRI-guided stereotactic localization, however, provides better multiplanar target resolution than is available with computed tomographic (CT) scanning. Therefore, we compared the accuracy of stereotactic coordinates determined by MRI and CT studies in 41 patients (53 targets). Coordinates were measured in each plane and as vector distances between the target and the center of the stereotactic frame on axial or coronal MRI studies. Absolute axial plane MRI and CT distances varied an average of 2.13 +/- 1.59 mm. The mean difference in measurements in the X (left-right) dimension was 1.19 mm and 1.55 mm in the Y (anterior-posterior) dimension. Central targets (located less than 2 cm from the frame center) had a mean MRI-CT difference of 2.09 +/- 1.79 mm; peripheral targets (greater than 2 cm from the frame center) differed by 2.17 +/- 1.3 mm. The voxel volumes were calculated for all compared images. Although differences between the physical properties of data acquisition with each imaging modality could explain the observed CT-MRI discrepancies, a 1-pixel difference in target selection could account totally for all the variance observed. MRI field strength (0.5 vs. 1.5 T) did not correlate with coordinate determination accuracy. We conclude that MRI-guided stereotactic localization can be used with confidence for most diagnostic, functional, and therapeutic stereotactic procedures.

MR imaging of normal rat brain at 0.35 T and correlated histology

A custom-built small-animal transceiver was used for in vivo imaging of normal rat brain at 0.35 T, with the objective of identifying anatomic components by comparison of images with corresponding histologic sections. The cerebrum, cerebellum, brain stem, ventricles, hippocampus, and subarachnoid space were identified and cerebrospinal fluid (CSF) was differentiated from gray matter and white matter on coronal and transaxial magnetic resonance (MR) images. These images compare favorably with those obtained by others at higher field strengths in regard to delineating major neuroanatomic structures. It is concluded that this technique will be useful for investigating small-animal models of human neurologic disease involving morphologic and morphometric changes in gray matter, white matter, and CSF-filled spaces.