Biomedical Engineering Reference
In-Depth Information
since measuring the water PRF directly can potentially result in
higher sensitivity given appropriate SNR with adequate spectral
analysis techniques (McDannold et al. 2001; Taylor et al. 2008).
The main disadvantages for CSI methods have been low spatial
and temporal resolution. Therefore, researchers have worked to
develop CSI techniques that increase the sensitivity and spatio-
temporal resolution while maintaining the ability to reduce arti-
facts seen in CPD.
Magnetic resonance spectroscopic imaging (MRSI) was
proposed for MRTI before the CPD technique was introduced
(Kuroda et al. 1996). However, the acquisition time needed for
the desired spectral resolution (with or without suppression) was
deemed impracticable for MRTI. For example, MRSI measure-
ments using a small 32 × 32 matrix with a spectral bandwidth
of 10 ppm at 1.5T and a spectral resolution of 0.01 ppm (1°C)
required 26 minutes per acquisition.
One approach to overcoming some of the time limitations of
the MRSI approach is to use echo planar spectroscopic imag-
ing (EPSI) techniques. These techniques have been proposed for
temperature monitoring, and prior studies have demonstrated
that temperature changes can successfully be monitored, lipid
can be used as an internal reference when present, and artifacts
due to motion can be reduced (Kuroda et al. 2000). Although
this method was demonstrated to be considerably faster than
conventional MRSI, initial EPSI acquisitions still required
3 minutes to acquire a low spatial resolution temperature image
(5 × 5 mm 2 ) on a 3.0T system. This is because the technique used
4 interleaved shots of 16 echoes (5.2 ms echo-spacing, total of
64 echoes) to achieve an effective echo-spacing of 1.3 ms to avoid
aliasing of the lipid signal due to a narrow spectral bandwidth.
Data interleaving resulted in a spectral bandwidth of 769 Hz
(12.04 ppm) with a spectral resolution of 12.0 Hz (0.188 ppm).
In addition, spectra from the EPSI technique can be degraded by
instability in the magnetic field and motion during the phase-
encoding process of the acquisition (Kuroda 2005).
McDannold et al. applied a line scan EPSI (LSEPSI) technique
(Mulkern et al. 1997) to address limitations to previous CSI-
based MRTI techniques (McDannold et al. 2001; McDannold et
al. 2007). LSEPSI is a combination of EPSI with a voxel-selective
technique for column scanning (Mulkern et al. 1997). This
method substantially improved the spatial and temporal resolu-
tion. Specifically, in an in vivo breast scan at 1.5T, a 32 × 32 cm 2
FOV was acquired in 4.2 seconds with a spatial resolution of 5 ×
5 × 5 mm 3 .
More recently, model-based approaches of multi-gradient
echo acquisitions have been investigated to encode the chemi-
cal shift of multiple chemical species such as water, methyl, and
methylene (Taylor et al. 2008; Li et al. 2009; Taylor et al. 2009;
Sprinkhuizen et al. 2010). These techniques hold potential to
overcome artifacts commonly seen with CPD techniques, such
as lipid contamination, to provide highly accurate and precise
thermometry at sufficient resolutions for even rapid thermal
ablations. One of the disadvantages is the increased computa-
tion time compared to CPD techniques. To address this, parallel
implementation on a stand-alone or portable workstation, which
can be easily adapted to the MR scanner, is now possible. For
example, implementation of the algorithms on a single graphics
processing unit (GPU) can, at best, achieve a speed-up factor of
125 (i.e., reduce the processing time by a factor of 125).
3.4 Summary
Invasive thermometry techniques are limited in their ability to
sample and complicate thermal therapy procedures. Techniques
utilizing invasive probes have the benefit of providing accurate,
real-time measurements in a specified location of interest, but
also the obvious limitations of being time consuming to localize,
increasing the risk of complications, and, depending on the pro-
cedure, may not be able to provide the feedback needed to ensure
the proper level of safety and efficacy desired. Bioheat models
can be used for limited predictions of temperature beyond the
probes but still do not provide accurate enough feedback for
many thermal therapies.
MRTI provides a means of noninvasive thermometry that
has facilitated the ability to perform some thermal therapy pro-
cedures that would otherwise not be feasible, as well as poten-
tially increasing the safety and efficacy of procedures that can
be performed in those environments. Current techniques for
MRTI have been incorporated into thermal ablation and hyper-
thermia protocols at numerous sites, and commercial products
for performing thermometry are beginning to get FDA clear-
ance from MR and third-party vendors. The next generation of
these techniques will have basic motion and field drift correc-
tion capability built into the systems. As with any thermometry
technique, implementation of quality control procedures for
evaluating the accuracy of MRTI in a relevant phantom with the
heating modality to be used clinically is critical. Additionally,
using this same equipment, verification of the temperature-sen-
sitivity coefficient should be considered using ex vivo samples of
relevant tissue when possible.
Despite more than a decade of progress, there are numerous
challenges and hurdles that remain for robust implementation
of the PRF technique. Attempts to mitigate PRF errors arising
from tissue motion have been addressed from both the acquisi-
tion and postprocessing side, but these techniques are still not
fully robust on their own. Additionally, susceptibility and sig-
nal loss from equipment and heating can complicate monitoring
near interstitially placed applicators. Potential solutions to these
problems are currently under investigation and may include
incorporation of multi-parametric data, such as diffusion (Das
et al. 2005) or T 1 (Ong et al. 2003; Taylor et al. 2009), to comple-
ment the PRF measurements. Additionally, the incorporation
of real-time modeling and simulation utilize MRTI feedback to
provide more optimal temperature estimates and extrapolations
into regions with errors or lacking signal. In all cases, the tech-
niques being investigated are making use of the steady advances
seen in high-performance computing in order to apply more
rigorous correction schemes in real time. This is a phenomenon
observable throughout medical imaging.
Search WWH ::




Custom Search