We read Dr. Ross’s editorial “The-High-Field-Strength Curmudgeon” (1) with great interest, particularly because we have had a very different—in fact, almost opposite—experience with brain imaging with our most-recent 3T system. Our group just completed its first year of 3T brain imaging and have also had 9 months of experience with imaging other parts of the body, including spine, head and neck, abdomen, pelvis, and all orthopedic applications. We are currently examining more than 25 patients a day with our new 3T MR imaging system. Approximately 65% of these examinations are of the brain.
Compared with the results obtained on current 1.5T systems with the strongest gradients, 3T has improved the quality of brain MR imaging at our institution (Figs 1 and 2). We have, however, taken a somewhat different approach to clinical imaging of the brain at 3T than that described by Dr. Ross. We did not seek to duplicate the same quality of images at 1.5T with increased throughput. Instead, we chose to pursue better spatial resolution at the increased field strength by using thinner sections (1–3 mm) with high matrices (256 × 256, and often higher). We perform multiple pulse sequences (T1-weighted, fluid-attenuated inversion recovery [FLAIR], T2-weighted, diffusion-weighted, and echo planar T2-weighted) and reformat the volume acquisitions in multiple planes both before (T1 and FLAIR) and after (T1 only) contrast enhancement. Our examination time is 30 minutes, which includes pre-imaging time. We have circumvented the problem of prolonged T1 relaxation at 3T (obscuring gray matter–white matter differentiation on spin-echo [SE] images) by using, as Dr. Ross alluded to, 3D T1-weighted spoiled gradient echo sequences (FSPGR/MPRAGE) obtained with 1-mm isotropic voxels with 256 × 256 matrices. The signal intensity disparity between gray and white matter is significantly greater by using this pulse sequence than the SE sequence at any field strength (Fig 3).
We perform this sequence in the axial plane except when we are imaging pituitary glands or patients with seizures whom we image in the coronal or off-coronal (angled perpendicular to the hippocampus) plane. The enhanced conspicuity of gadolinium at 3T over 1.5T obviates the old, but still controversial, argument that only SE is adequate for detecting disease on postcontrast images. We have also had two separate manufacturers create 3D FLAIR fast SE (FSE) sequences with 1-mm isotropic voxels using 256 × 256 matrices that we acquire in the sagittal plane (Fig 4) and reformat in the axial and coronal planes. For seizure patients, we acquire this pulse sequence in the off-coronal plane angled perpendicular to the hippocampus. An added benefit we observed in obtaining FLAIR as a volume acquisition is the dampening of increased CSF flow artifacts one invariably sees with 2D FLAIR pulse sequences at 3T. As Dr. Ross commented in his editorial, high-spatial-resolution FSE T2-weighted images are a strength at 3T. We currently are using a T2-weighted FSE sequence with 512 × 384 matrices and 3–4-mm section thicknesses (Fig 2) and are awaiting the completion of a 3D T2-weighted FSE sequences with 1-mm isotropic voxels (256 × 256 matrices), which we requested from the manufacturer. With the advent of fast FLAIR imaging, we have not used balanced imaging in the brain in for nearly 5 years.
Radiologists seeking to push the envelope with 3T should not despair. In reference to specific absorption rate (SAR), the new whole-body 3T MR imaging system that we have been using for 4 months has not had a single SAR error with brain imaging, and there have been very few power deposition problems while imaging other parts of the body. The increase in chemical shift artifact at 3T has not had a deleterious effect on our ability to evaluate disease if adequate band widths are used, and although magnetic susceptibility artifacts are exacerbated at 3T (2), they can be mitigated by increasing band width and echo train length, decreasing TE and section thickness, and aligning the frequency encoding direction parallel to the long axis of any metal (Figs 5, 6, and 7). When a relatively new technique for multishot SE with radial orientation of k space becomes available for 3T, this problem will be further diminished (3). We are in agreement with Dr. Ross’s and others’ (4) opinions that MR angiograms are better at the higher field strength. The added signal-to-noise benefit at 3T has significantly improved our diffusion-weighted and apparent diffusion coefficient images, and that factor plus the advantage of increased magnetic susceptibility effects at 3T, has also made our perfusion studies better. Similar to Dr. Ross, our initial impression with 3D spectroscopy is that we have not yet seen a significant benefit over 1.5T.
Our experience with 3T MR brain imaging during the past 4 months in over 1000 patients has benefited from our new, second generation whole-body MR imaging system. By using the pulse sequences 3D T1-weighted FSPGR without and those with contrast medium administration and 3D FLAIR (both obtained with 1-mm isotropic voxels, 256 × 256 and 220 matrices), T2-weighted FSE with 3-mm section thickness, 512 × 384 matrices, as well as diffusion and echo planar T2-weighted images, we are able to perform brain imaging in 30 minutes with quality we judge tobe superior to that obtained with current 1.5T systems.
5 mm Axial T2 TSE obtained on 1.5T in a patient with tuberous scvlerosis showing equivocal lesion (arrow tip) adjacent to left foramen of Monro.
4 mm Axial T2 FSE, 512 × 384, obtained on 3T 12 weeks later in same patient as Figure 1 with arrow tip on more obvious lesin adjacent to left foramen of Monro. Note increased flow artifacts in phase direction which are exacerbated by 3T. These will be significantly diminished with multidirectional flow comp or with a 3D T2FSE acquisition.
1 mm Direct Sagittal FLAIR FSE, 256 × 256, demonstrating the only lesions (3mm subependymal nodule [white arrow] and a 2mm subcortical tuber [black arrow]) in another patient with tuberous sclerosis.
1 mm Reformatted Coronal T1 FSPGR from same patient as Figure 3 obtained in the axial plane with 1mm isotropic voxels, 256 × 256, revealing the subependymal nodule (black arrow) in the superolateral aspect of the left lateral ventricle.
2 mm Sagittal T1 FSE obtained on 3T showing adequate cord visualization (in patient with a previous anterior fusion) by varying 5 parameters (increasing bandwidth and echo train length [ETL], decreasing slice thickness and TE, and orienting frequency encoding gradient parallel to long axis of metal).
2 mm Sagittal T2 FSE at 3T from same patient as Figure 5 demonstrating good visualization of the spinal cord by using the aforementionted techniques.
3 mm T2 Axial T2 FSE with fat sat through the C6-7 foramina, demonstrates adequate visualization of cord and nerve roots (except proximal left C7). If this acquisition had been obtained with 2 mm slice thickness as well as maximum bandwidth and ETL and the lowest TE, the left C7 root may have been seen in its entirety.
References
Reply
I thank Drs. Shapiro, Magee, Williams, Ramnath, and Tannenbaum for their interest in my editorial and for their detailed responses regarding their usage patterns, sequences, and experience at 3T. The rapid pace of technological advancement at 3T makes purchase decisions even more difficult, compounded by the usual differences in manufacturers, imaging hardware, software, and postprocessing capabilities. Their letter further demonstrates the point—which I tried to make—that moving from 1.5T to 3T is not trivial. These sophisticated users have shown that good image quality can be obtained at 3T, but the old adage remains true now more than ever—caveat emptor. To this I would add caveat vendor, let the seller beware. The responsibility of quality image production is not a one-way street, and considerable burden is placed upon the manufacturers not to bring systems to a general clinical use market before real-life imaging demands can be met.
- Copyright © American Society of Neuroradiology
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.