3T Neuropediatric Imaging, Initial Experience at the Children's Hospital of Philadelphia
Robert A. Zimmerman1, Heiko Meyer2
1Children's Hospital of Philadelphia, Philadelphia, USA; 2Siemens Medical Solutions, USA.
| May 31, 2007
Six months ago a new 3T MAGNETOM Trio was installed at Children’s Hospital of Philadelphia. After three months operating from 8am to 4pm mostly on out-patients, the system was then utilized from 8am to 10pm in line with our standard clinical hours of operation. The initial three months were used to focus on protocol development and comparisons to the 1.5T systems MAGNETOM Avanto and MAGNETOM Sonata also set up in our hospital. In the meantime we have scanned over 800 patients, more than 500 of them being sedated.
On the one hand, since the relaxation time T1 increases, while T2 decreases, when moving from 1.5T to 3T, we anticipated that some protocol optimizations would be needed to maintain a certain contrast. On the other hand, 3T offered the opportunity to use the higher signal-to-noise ratio (SNR), the longer T1 times, and the greater contrast medium (Gadolinium) conspicuity to improve the diagnostic quality. The question was, however, whether routine studies on the MAGNETOM Trio could be done with sufficient diagnostic accuracy and satisfactory throughput to substitute for another 1.5T system. To address this question, we put the same load on the MAGNETOM Trio as on the other clinical scanners, working on a one-hour time slot per patient basis. Then we developed protocols to use the additional SNR for protocols with higher resolution and shorter scan times.
Fig. 1 8-year-old male with Neurofibromatosis, 1 mm isotropic 3D TSE T2.
The ability to use iPAT in numerous protocols proved to be vital to achieve the acquisition of more data in the same or a shorter period of time. A 3D T2-weighted non-iPAT scan covering the whole brain takes 9:25 min with a matrix size of 256x192, while the scan time for an isotropic matrix of 320x320 can be reduced to 4:30 min with an iPAT factor of 2. The results can be seen in Fig.1.
Although T1-weighted scans are in general faster, the application of iPAT allowed us to shorten a whole head 3D acquisition with a 256x244 matrix from 7:05 min to 3:56 min. The grey-white matter contrast using the 3D MPRAGE sequence is excellent, providing the ability to reconstruct images with sub-millimeter resolution.
Fig. 2 3D MPRAGE of an 11-year-old male post operative tumor resection of an dysembryoplastic neuroepithelial tumor with residual tumor.
Fig. 2 shows an example of an 11- year-old male who underwent surgery for DNET tumor resection. The images clearly show residual tumor in the cavity, which was not enhancing and remained elusive on 12 previous follow up MRI studies on a 1.5T system over a period of 31/ 2 years. The ability to demonstrate the tumor unequivocally depended on the 3T higher field strength and signal-to-noise that allowed excellent quality thin section resolution of the surgical cavity margins, not achievable at 1.5 Tesla.
Fig. 3 1.5T MRA (left) and 3T MRA (right) of a 4-year-old male with absent left internal carotid artery.
3D techniques to acquire isotropic voxels allow reformatting the images in any given plane, as well as with different slice thicknesses, which we use to obtain transverse and coronal views. This way we avoid scanning one or two additional planes, which helps to save valuable time.
Time-of-Flight Magnetic Resonance Angiographies (TOF-MRAs) also benefit from the higher SNR, as well as from the longer T1 relaxation times, which allow the imaging of smaller and deeper vessels. Where on a 1.5T system the acquisition times are in the range of 9 to 10 minutes, the use of iPAT allowed the acquisition of the same 512x224 matrix in 4:16 min. Smaller and deeper vessels can be depicted as can be seen in Fig. 3.
Fig. 4 3T Arterial Spin Labeling image of a 4-year-old male with absent left internal carotid artery.
Another application, for which the longer T1 times at 3T are beneficial, is Arterial Spin Labeling (ASL). This technique utilizes the blood as an intrinsic contrast agent by labeling the inflowing blood and using it to measure perfusion. The longer the T1, the deeper inside the brain perfusion can be measured. Since the signal difference between unlabeled tissue and labeled blood is less than 10 percent , SNR is also a crucial point. We imaged the above mentioned 4-year-old with a home-built ASL at 3T, which shows that both hemispheres are perfused despite the absent left internal carotid artery (see Fig. 4). Compared to contrast enhanced perfusion imaging, ASL has the advantage that measurements can be easily repeated if, for example, there is patient motion degrading the diagnostic quality.
Fig. 5 Diffusion-weighted image with b = 1000 s/mm2 of a 14-month-old male with sickle cell disease after acute infarction.
When looking at acute infarction, diffusion weighted imaging is one of the key techniques to asses the tissue at risk at an early stage. Fig. 5 shows a diffusion weighted image with a b-value of 1000 s/mm2 using a 192x192 matrix, which reveals the infarction area.
Alongside the advantages of 3T, we were aware that a higher field strength would increase the local radio-frequency power deposition (SAR) and the risk potential for medical devices and surgical implants. Therefore we only admitted patients with implants to the 3T scanner, which were explicitly compatible with a field strength of 3T. To counteract the higher SAR, especially for T2-weighted sequences, we routinely run the SPACE (Sampling Perfection with Application optimized Contrast using different flip angle Evolutions) sequence which can reduce the SAR by up to 80%.
SPACE is using a Turbo Spin Echo type of acquisition scheme, but instead of sampling 15–25 echoes, as it is usually done for T2-weighted imaging, 200–400 echoes are acquired in each echo train. To compensate for the strong signal decay that would normally occur with this HASTE-type of acquisition, the flip angle of each refocusing radio-frequency pulse is specifically calculated and modified. This maximizes the contrast between different tissue types, e.g. gray and white matter and CSF and it also avoids the blurring due to T2 decay, which occurs in HASTE-type of scans with such a long echo train. The long echo train made it feasible to cut down the scan time for an isotropic whole brain scan to 4–6 minutes, while providing a high degree of details in all planes. Although an in-plane resolution of ~1x1 mm2 was standard at our institution, we typically acquired 3–5 mm thick sections, making it necessary to run separate scans for different orientations.
Fig. 6 Susceptibility-weighted scan of a 1.5-months-old infant.
Using the SPACE technique we now scan in one orientation and reconstruct the other planes maintaining the high resolution. The higher sensitivity to susceptibility differences in the tissue at 3T is helpful when looking at intracranial hemorrhages (see Fig. 6), but it also degrades the image quality in the presence of e.g. braces. However, sequences with higher bandwidth and shorter echo spacing, as well as the use of iPAT can reduce this effect.
In conclusion, we found that the MAGNETOM Trio is able to handle the full clinical load, providing in some cases superior diagnostic information compared to 1.5T imaging. Further improvements will involve protocol optimizations to reduce susceptibility-induced artifacts and setting up standards for metallic implants. Due to the extended use of high-resolution 3D imaging techniques, new strategies need to be developed for reading the substantially increased amount of images.
We are planning a Tim upgrade this fall which should increase the performance of the system even further and give us the same technology as on the MAGNETOM Avanto. One of the reasons to upgrade is the ability to then use iPAT in any direction in any region of the body. The 12-channel Head Matrix coil will allow iPAT factors of 4–6 for standard applications whereas the Tim concept of using as many coil elements as necessary for a specific examination will improve patient throughput and diagnostic quality even more. In a pediatric environment, the size and the weight of the coils are always of concern and we see a clear advantage of the Tim technology which features adequately sized coils, while providing high SNR in conjunction with lightweight coils. Not having to change the coils for combined head and whole spine examinations, for example, will be beneficial given the already long sedation time for this kind of study and the danger of the child waking up during repositioning. All this has a positive effect on the safety of the patient and efficiency of the scan. After all upgrades are done we will have five state-of-the-art MR systems to provide exceptional patient care as well as a platform for research studies – reflecting our commitment to remaining one of the leading pediatric hospitals worldwide.