SEATTLE - Like true Seattleites, nearly a thousand visiting radiologists mostly ignored the Sunday morning downpour to carry on with business. Their business was the 101st annual meeting of the American Roentgen Ray Society, which began at 8:30 sharp in a packed meeting room in the city's convention center.
"I expected to be more like the warm-up band ... so I really do appreciate this turnout," said a slightly surprised Dr. Kenneth Buckwalter, associate professor of radiology at the Indiana University School of Medicine in Indianapolis. Buckwalter began by carefully comparing Seattle's weather forecast for the coming week (more rain) to that of Indianapolis (less rain), then moved quickly to the drier subject of MRI protocols in musculoskeletal imaging.
He began with some basics. The magnet's field strength is fixed, he said, and the noise level generated by different magnet strengths is also fairly constant. Higher field strengths increase the signal-to-noise (SNR) ratio, which improves the ability to detect low-contrast structures. So higher field strength is better, but not always.
"Since the signal's reduced, and correspondingly, the SNR is reduced at mid and lower field strengths, is there anything good about imaging at lower field strengths?" Buckwalter asked.
There are at least two good things. High field strengths increase the incidence of both chemical-shift artifact and its sort-of cousin, magnetic-susceptibility artifact, both of which can lead to misdiagnosis.
"Chemical-shift artifacts are due to the fact that water and fat resonate at slightly different frequencies," Buckwalter said. "Frequency determines the spatial location of the object that we image. When imaging muscle, cartilage or fluid surrounded by fat, we will see a white stripe at one boundary and a black stripe at the other boundary, which is the interface of tissue or water and fat. In fact, I've seen people call this cartilage thinning in the knee, when in reality it's a chemical-shift artifact. You need to be very careful about this."
To look for magnetic susceptibility artifact, look at the bone marrow, he said. When magnetic susceptibility in high-field-strength scans renders the marrow black, pathology anywhere can lose contrast and be harder to diagnose.
On the other hand, low- to mid-field-strength magnets have the built-in disadvantage of lower SNR. To compensate for lower-field-strength magnets, radiologists can deploy a variety of strategies that increase SNR significantly by compromising in other areas, such as reducing spatial resolution or enlarging the field of view.
"The natural response for most of us is to double or triple or quadruple the [scan time]," Buckwalter said. "Probably a more reasonable strategy is to reduce the phase-encoding. In others words, take a slight hit in the spatial resolution. It will decrease the scan time and it will decrease the resolution, but it will increase the SNR. Again, make it just enough to carry you over that threshold where you'll be able to see the pathology. By thicker slices, I'm not talking about going from 3 mm to 10 mm, I'm talking about going 3 to 4, or 4 to 4.5 [mm]. Larger fields of view -- very similar. No effect on scan time, decreasing resolution."
Coils and positioning
One of the easiest ways to increase signal intensity without decreasing spatial resolution is to use a dedicated joint coil, he said. Because signal strength is proportional to the coil's radius cubed, a small change in the radius of the coil vastly increases the available signal strength.
The coil must also be positioned properly, he said. Coil positioning relative to the object being imaged follows an inverse square law -- not as important as the cubed law that governs coil radius, mind you, but important. In knee imaging, the center of the coil should be about one or two fingers below the patella. Proper positioning is especially important with lower-field-strength magnets, and the initial scout image can determine if the coil needs repositioning.
Don't be greedy
"The second thing is, don't be greedy. In other words, reduce the spatial resolution just a little bit, use a slightly thicker slice and a slightly larger field of view," Buckwalter said. "Don't try to fit a 1.5-tesla protocol on a 0.3[-tesla] magnet. Try to do that in lieu of increasing the number of excitations."
This part can even be done from the comfort of home. The Indiana University group has an SNR calculator on its Web site, specifically at http://www.indyrad.iupui.edu/public/kbuck/snr, where plugging in a few numbers shows the effects on SNR of small changes in field of view, slice thickness, imaging matrix, and other variables, Buckwalter said.
It's also important to orient the frequency-encoding direction perpendicular to joint surfaces, he said. Conventional MRI images consist of a frequency-encoding and a phase-encoding direction. Acquiring fewer than the maximum number of encoding steps degrades resolution in the phase-encoding direction, while placing the frequency-encoding direction perpendicular to the articular surfaces optimizes joint visualization. While this step can increase flow and motion artifacts, they are usually not significant enough to prevent interpretation.
"So we can turn lemons into lemonade, particularly at low field strength, particularly if we understand a little bit about this," Buckwalter said.
Use fast spin-echo and inversion recovery
Fast spin-echo techniques usually provide excellent T2-weighted or STIR (short T1 inversion recovery) imaging, he said. They allow multiple signal averages with reduced scan times.
The STIR technique is more reliable than fat-suppression techniques, Buckwalter said. An abstract accompanying his presentation noted that STIR is more selective than fat suppression with frequency-selective radiofrequency pulses (fatsat) because inversion recovery does not depend on a homogenous main magnetic field. The STIR technique also works in the presence of metal, and is the only fat-suppression technique available on most mid- and low-field-strength systems, according to the abstract.
Use reliable protocols and uncouple scans
"As I supervise a number of different scanners that are remote from my location, I cannot view in real time every scan," Buckwalter said. "I need to know that the sequences I select will work every time, and I can tell you that fatsat does not work all the time."
In order to save time and avoid having to redo scans, radiologists should consider routinely acquiring two kinds of images with two protocols: one that shows the anatomy and one for the pathology, he said.
"It's best to avoid multiecho proton-weighted density and T2-weighted pair. This technique requires long scanning times, and if the patient moves, the whole thing has to be redone," Buckwalter said. "If you uncouple the scans and have two shorter scans, patient motion typically affects only one of these sequences, and it’s easy to repeat. Furthermore, you can optimize each scan sequence separately, and often there's no time penalty for doing this; there may be potential time savings."
Gradient-echo imaging in high-field systems
Finally, advances in gradient-echo imaging have enabled short TE times for gradient-echo imaging, which can be a powerful tool for use in high-field scanners, according to Buckwalter, who uses a Signa LX 1.5-tesla scanner (GE Medical Systems, Waukesha, WI) at his institution. The techniques, developed for use in breath-hold body imaging, can be very successfully applied to musculoskeletal MRI, he said. One such technique, FMPSPGR (fast multiplanar spoiled gradient), is used as a substitute for a T1-weighted spin-echo sequence in order to reduce acquisition time.
"If you use the absolute minimum TE to remain 'in-phase,' you can produce a very respectable T1-weighted image in a minute in a half," Buckwalter said. "We've got the equipment, there's no reason we shouldn't take advantage of this. We can do a lot more, faster and better, with no patient motion."
By Eric Barnes
AuntMinnie.com staff writer
April 30, 2001
Click here to post your comments about this story. Please include the headline of the article in your message.
Copyright © 2001 AuntMinnie.com
