Recent developments have led the move of 3T magnetic resonance imaging (MRI) from a research setting to clinical practice. This article reviews the use of these high-field scanners to get high-quality diagnostic images for a variety of clinical indications. With details on technical parameters and settings, the author recommends how to optimize the use of 3T MRI to utilize the additional signal it provides.
Over the last several years, systems operating at higher fields
have become more prevalent, particularly at research centers. An
informal survey of the market reveals that approximately 250
whole-body-capable MR systems are currently in operation with the
majority of recent installations in the clinical practice setting.
Market data in late 2004 indicate that 3T systems make up 25% of
new high-field MR purchases.
Fueling the shift in interest from 1.5 to 3T from primarily a
research device to clinical practice validates that what was once
considered very high-field MR (3T) is practical, feasible, and,
indeed, currently or potentially superior to 1.5T for clinical
indications throughout the body. The driving forces behind this
increased penetration of 3T scanners into the clinical setting
include reduced concerns over surface coil availability,
radiofrequency (RF) deposition limit, higher ambient noise, system
homogeneity, increased magnetic susceptibility and chemical shift
effects, and reduced tissue contrast. Also, this shift results from
the documentation of incremental benefits of 3T over 1.5T with
respect to image quality and efficiency.
Specific absorption rate
Specific absorption rate (SAR) is a measure of energy deposited
by an RF field in a given mass of tissue. SAR is established by the
International Electrotechnical Commission (IEC) to not exceed 8
watts per kg (W/kg) of tissue for any 5-minute period or 4 W/kg for
a whole body averaged over 15 minutes.
Dissipation of RF energy in the body can result in tissue heating.
The doubling of field from 1.5T to 3T leads to a quadrupling of SAR
(Figure 1). Therefore, SAR considerations effectively limit scanner
Manipulations traditionally used to limit SAR include reducing
acquisition flip angle (eg, from 180˚ on fast spin-echo [FSE] and
~40˚ on gradient-recalled echo [GRE]), which could potentially
affect image contrast. Longer echo time acquisitions commonly used
with high-per-formance gradient systems and fat-suppression
techniques commonly used with musculoskeletal imaging exacerbate
the duty cycle load. Reducing duty cycle by using a longer
repetition time (TR) than the minimum necessary (building in
cooling time, in a sense) is an effective technique but comes at
the expense of somewhat longer scan times (Figure 2). Parallel
imaging is another powerful method of reducing RF exposure as well
as scan times by reducing the number of phase-encoding steps that
are performed. The typical trade-off in signal-to-noise ratio (SNR)
(a parallel imaging factor of 2 reduces SNR by 40%) is balanced by
improved surface coil design and the higher signal acquired at 3T.
Another technique of managing SAR concerns is to interleave
SAR-intensive sequences with low RF deposition scans, eg, follow a
fat-suppressed FSE scan with a gradient-echo acquisition before
starting the next FSE scan.
Innovative methods of reducing SAR without image compromise are,
or will soon be, available. New magnet designs now seen in the
clinical setting are inherently more SAR-efficient than were
earlier generation systems. Clever pulse sequence manipulations
such as applying magnetization transfer prepulses only at the
center one third of k-space can maintain improved tissue contrast
while depositing considerably less RF energy. Advances in pulse
sequence design such as reshaping RF and gradient waveforms reduces
peak RF power up to 40% compared with conventional techniques. The
development and availability of more local transmit/receive surface
coils will also reduce SAR deposition and further enhance
Sound pressure levels (SPLs) increase with field strength. The
noise levels at 3T approach twice that of 1.5T and can be in excess
of 130 dBA
(the IEC and the U.S. Food and Drug Administration limit
permissible sound levels to 99 dBA). Higher gradient performance
comes at the cost of higher SPL as well. Magnet length also
influences the gradient noise generated, thus the shorter bore
systems sold today are inherently louder.
Methods of reducing SPL include passive approaches, such as the
routine use of earplugs, as well as active noise cancellation via
headphones. Reduced gradient performance for certain applications
is another approach, but by nature, this limits clinical efficacy.
Some currently available 3T systems are equipped with advances,
such as acoustically shielded vacuum-based bore liners that keep
noise levels below certain limits while maintaining full gradient
Tissue contrast issues
T1 relaxation times are prolonged at 3T with respect to 1.5T
leading to reduced contrast resolution on traditional (short TR,
short echo time [TE]) spin-echo (SE) acquisitions. These
considerations do not plague other methods for obtaining T1
contrast, such as RF-spoiled gra-dient-recalled, magnetization
prepared techniques such as inversion recovery (IR), or
magnetization transfer (MT) 3-dimensional spoiled
gra-dient-recalled (SPGR) (Figures 3 and 4). Inversion recovery
techniques that produce superior T1 contrast at 1.5T, such as
phase-sensitive IR for the brain (Figure 5) and T1 fluid attenuated
inversion recovery (FLAIR) for the brain, spine, and
musculoskeletal system, are equally well suited to higher field
imaging and can yield spectacular results. With parallel imaging
techniques, T1 studies are faster and higher in resolution than
those obtained at 1.5T. A routine shift to high bandwidth, to a
moderate echo-train (ET)-FSE (T1 FLAIR) from spin-echo, has the
additional benefit of reducing susceptibility artifact, which is a
benefit in patients who have had surgery or who have metal
implants, and chemical shift effect sensitivity as well.
While the relaxivity of gadolinium is not significantly
different at 1.5T from at 3T, the longer T1 of tissues at 3T
contributes to an increase in conspicuity of enhancement (greater
contrast-to-background ratio). Therefore, many sites utilize a
lower dose of contrast (0.05 mmol/kg) for routine brain imaging
purposes. T2 values in biological tissues are unchanged or only
slightly decreased with increases in field strength. T2* effects
scale with field strength, and 3T studies are thus more sensitive
to deposition of blood products and tissue mineralization (Figures
6 and 7). Conversely, susceptibility artifacts are proportionally
more problematic at 3T (Figures 8 and 9). The higher SNR afforded
by 3T augmented by the power of the latest generation phased-array
coils allows a variety of techniques to compensate for T2* effects,
including the use of parallel imaging and higher bandwidth with
longer ET-FSE acquisitions.
The greater signal intensity afforded at 3T is particularly
enticing for diffusion-weighted imaging (DWI) needs.
Signal-to-noise ratio can be marginal for routine clinical imaging
purposes at 1.5T, and the quest for higher B values (>1000 s/mm
), thinner slices (<3 mm), and white matter anisotropy mapping
(tensor imaging) further stresses the SNR equation (Figure 10).
Diffusion-weighted imaging studies at high field are typically
acquired using echoplanar imaging (EPI) techniques. These
single-shot studies are inherently prone to susceptibility
artifact, which can limit evaluation of structures in close
proximity to the bony skull base and air-filled paranasal sinuses.
Since susceptibility effects scale with field strength, these
artifacts are proportionally worse at higher field. Parallel
imaging techniques are routinely applied on modern 3T systems
equipped with optimized surface coils and broadband reconstruction
hardware, effectively balancing these considerations by decreasing
the echo-spacing (ES) and TE of the scan. This reduces
susceptibility artifact and ameliorates signal loss due to T2 decay
on these long ET acquisitions (Figure 11). When widely available,
less artifact-prone FSE techniques (eg, Propeller) to obtain DWI
should become popular at 3T (Figure 12).
Perhaps the greatest neuroimaging impact of 3T is in enhancing
the capability of functional MR imaging (fMRI). The greater
susceptibility, contrast sensitivity, and higher SNR inherent to 3T
scanning can produce up to a 40% increase in detected activation
with blood-oxygenation-level-dependent (BOLD) imaging over 1.5T.
Improved contrast resolution enhances the success rate of these
procedures for routine presurgical mapping of eloquent cortex (eg,
sensorimotor, language) and coupled with scanner integrated
paradigm delivery may lead to utilization in community practice to
evaluate disorders such as dementia and psychiatric disorders
Time-of-flight neuro MR angiography
The longer T1 of background tissues can be exploited for
superior inflow MR angiography (TOF MRA). Scanning techniques
employ lower flip angles that reduce SAR deposition as well as
pulsation artifacts. The higher SNR provided by 3T with 8-channel
surface coils encourages routine utilization of high imaging
matrices (512 × 1024) producing studies that can rival the
resolution of digital subtraction angiography (DSA) (Figures 14 and
15). Optimized coils coupled with parallel imaging techniques
maintain scan times similar to or shorter than those at 1.5T.
Chemical shift doubles when moving from 1.5T to 3T, resulting in
improved spectral resolution allowing evaluation of metabolites
that may be obscured at 1.5T. This factor along with the higher SNR
of 3T may increase the efficacy of proton and multinuclear
spectroscopy of many disorders (Figure 16).
Specific absorption rate considerations reduce slices available
per given time, encouraging multiple breath-hold acquisitions.
Motion-resist-ant techniques with single-shot FSE and
respiratory-triggered multishot FSE are also commonly utilized.
Eight-channel phased-array surface coil designs optimized for
parallel imaging ameliorate many SAR-based limitations. Studies of
the abdomen and pelvis are routinely accomplished with thinner
slices and higher imaging matrices, comparable to those utilized
with computed tomography, facilitating comparison and lesion
characterization (Figures 17 and 18). The higher SNR afforded by 3T
may also facilitate applications such as spectroscopy and might
obviate the need for endocavitary surface coils for advanced
applications, such as prostate imaging (Figures 19 and 20).
The higher SNR of 3T coupled with parallel imaging compatible
surface coils produce high-quality, perhaps higher resolution,
contrast-enhanced vascular studies with greater consistency than at
1.5T. Lowering the flip angle reduces SAR, and thus acquisition
time. The longer T1 values of background tissues serves to augment
visualization of intravascular contrast, potentially allowing a
reduction in contrast dose administered (Figure 21). While
full-body vascular coils are yet to become available, the
increasing importance of multistation time-resolved MRA techniques
at the expense of so-called bolus-chasing reduces their
significance (Figure 22).
Eight-channel phased-array coils are widely available for spine
imaging. Practical considerations yield studies that are generally
higher in resolution and somewhat faster than at 1.5T (Figures 23
and 24). While susceptibility is a theoretical concern, long ET
high-bandwidth acquisitions yield excellent image quality even for
patients with implanted metal hardware.
Responsible for upwards of 20% of the study volume of the
typical clinical scanner, the quality of joint imaging is a major
factor in determining the financial feasibility of higher field MR.
Until recently, coil availability has been limited and SAR concerns
are prominent as high duty cycle applications, such as fat
suppression and long ET FSE, are common. The homogeneity of the
latest generation short-bore devices is critical as joints are
rarely scanned near isocenter and fat suppression is so important
for adequate contrast resolution.
Fortunately, high-quality, high SNR, often phased-array surface
coils are becoming available and are generally providing studies
that are recognizably superior to those from 1.5T systems in
somewhat less time (Figure 25). Offering higher spatial resolution,
3T MR may yield additional use in the study of smaller parts and
cartilage than do examinations obtained at 1.5T (Figure 26).
Studies obtained with older and less sophisticated coils (eg,
quadrature) benefit from the SNR of 3T to be competitive with
studies obtained with more advanced surface coils at 1.5T (Figure
Receive-only coils require SAR-intensive body coil transmission
that limits performance. The future availability of
transmit/receive capable surface coils should significantly augment
efficiency and further extend quality.
The higher SNR of 3T allows utilization of higher resolution
protocols with smaller fields-of-view (FOV), thinner slices, and
larger imaging matrices. The greater susceptibility sensitivity of
3T should make tissue mineralization easier to appreciate.
Instrumented joints can be imaged with manageable artifact with
high bandwidth, long ET-FSE, and T1-weighted IR-FSE techniques.
Fundamentally, 3T MR imaging offers twice the signal of 1.5T.
Specific absorption rate considerations are becoming less of a
limitation due to technical advances and surface coils that are
available for all core applications. The greater amount of signal
can be manipulated to make scanning more comfortable (scan times
that may be half as long) or more sensitive (higher resolution) to
smaller lesions (eg, small and early multiple sclerosis plaques).
Typical utilization combines both benefits in a given scan. The
greater sensitivity to magnetic susceptibility effects offers
unique benefits in functional neuroimaging as well as improving
sensitivity to brain hemorrhage. Practical considerations make
studies of the brain, spine, chest, abdomen, pelvis, as well as the
vasculature and extremities obtained at 3T consistently higher in
quality than those obtained at 1.5T.