This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
MRI is a noninvasive technique allowing for the visualization of structural and functional information involving the human body. The history of MRI can be divided into three parts: (1) the discovery of the NMR phenomenon (non-imaging), (2) the success of medical imaging with MRI (diagnostic, anatomic imaging), and (3) the emergence of functional MRI (fMRI, advanced imaging).
Isidor Isaac Rabi first described the NMR phenomenon in 1938 and developed a technique for measuring the magnetic characteristics of atomic nuclei. Not only was this discovery an important development in physics and chemistry, unforeseeable to Rabi, the discovery later facilitated the development of MRI for use in medicine. For his work with NMR, Isidor Rabi was awarded the 1944 Nobel Prize in Physics. While Rabi's work resulted in a molecular beam method for magnetic resonance detection, two other physicists, Felix Bloch and Edward Purcell,1,2 first observed, independently, NMR phenomenon in liquids and solids. The two shared the 1952 Nobel Prize in Physics for this discovery which further laid the physical basis for MRI. In 1951, Erwin L. Hahn developed a spin-echo method to study molecular diffusion in liquids3, and Richard R. Ernst developed Fourier transform NMR spectroscopy in 19664, winning the 1991 Nobel Prize in Chemistry for that work. It was not until 1971 that Raymond Damadian utilized NMR in biomedical applications, measuring T1 and T2 relaxation times rat tumors. Dr. Damadian observed that tumor tissue possessed longer T2 times than normal tissue, a finding which was published in Science.5 In that article, Dr. Damadian correctly predicted that the technique might "prove useful in the detection of malignant tumors". Paul C. Lauterbur and Peter Mansfield subsequently and independently described the use of magnetic field gradients to localize NMR signals in 1973, a technical development that laid the foundation for MRI as it is currently performed.6,7 Lauterbur and Mansfield shared the 2003 Nobel Prize in Physiology/Medicine for their contributions; Damadian was somewhat controversially excluded. The first MR images of humans were produced in 1977. On July 3rd 1977, Damadian's machine "Indomitable"-which is on display at the National Inventors Hall of Fame-produced crude images of the human thorax. Mansfield et al published an MR image of in vivo human anatomy-a cross-sectional image of the human finger-in a 1977 article in the British Journal of Radiology.8 Subsequently, imaging of the human wrist, abdomen, head, and body was performed from 1977 to 1980.9-12 Both Oxford Instruments, a British company spun-off from Oxford University, and FONAR, an American company founded by Dr. Damadian, claim production of the first commercial whole-body MR scanner in 1980. General Electric and Siemens entered the market shortly after and in 1983 produced their first commercial scanners. Fundamental work in diffusion imaging by Michael Moseley in 1984 laid the groundwork for functional MRI techniques,13,14 and in 1986, Le Bihan et al reported the calculation of diffusion coefficients using Moseley's method.15 In the following years, diffusion tensor imaging (DTI) and fMRI using blood oxygenation level-dependent (BOLD) techniques were also developed.16,17 Today, an ever-growing array of novel techniques for MRI is currently under investigation.18-20
Field Strength Evolution
The main magnet is an essential component of the MR unit, providing the static magnetic field and determining the performance of the entire MR system. The first commercial MR units became available in the early 1980s, and thereafter clinical systems at field strengths between 0.3 and 0.6T were widely installed.21-23 Note should be made that the original whole body MR system, developed at the University of Aberdeen, operated at 0.04T. Such low field MR systems commonly utilized permanent magnets or electromagnets to produce the main magnetic field (B0).24 Even today, low-field systems possess some small advantages over their high-field (1.5 and 3T) counterparts, including affordability, both in initial and operation costs, and fewer site-based constraints.25 Certain low field MR units also offer a simple open system design with the potential for improved safety26, leading to their use at times in the history of MR in interventional and intraoperative applications.27 Nevertheless, high field 1.5 and 3T magnets are the current clinical standards. These and ultra-high field unites utilize superconducting magnets to produce B0. 1.5T MR systems were initially introduced in 1985,28 and subsequently dominated the market as the gold standard for clinical imaging. In 1993, research institutions began investigating applications for 3T MRI and in 1998, 3T MRI was used clinically for brain imaging. Since then, whole-body 3T MR systems have been installed at innumerable sites worldwide, representing the standard of clinical care (when cost is not a consideration), with the clinical use of 3T continuing to dramatically increase.29 Even higher field MR systems (4-9.4T and higher) have been installed at research institutions in the last decade.30-32 The main motivation for higher field strength MR is the expected increase in the signal-to-noise ratio (SNR) and spectral resolution proportional to the field strength.33-36 Technical and economic factors somewhat limit this potential37, with physiological effects also of potential concern due to increased energy deposition and other factors. Thus far, only a small number of transient effects relating to the latter concerns have been reported.38,39 However, potential occupational hazards from exposure to MR scanners have recently been described.40-42 Future increases in field strength, for clinical purposes, will depend on actions of the Food and Drug Administration as well as the International Electro-technical Commission-an international commission monitoring standards for electrical devices.43 Today, dozens of clinical studies at 7T attest to the fact that ultra-high field strength imaging has become increasingly popular.20,44,45
Open MR systems
The concept of "open MRI" continues to evolve. Initially, the term referred to a open magnet with a C-arm or vertical design rather than a horizontal bore. Currently, the term "open MRI" typically refers to a wide, short horizontal bore design. Rothschild initially described a 0.064T open MR system in 1991 consisting of "a vertical field permanent magnet supported by four posts [and a] 42 cm gap between the table and the magnet".46 That system allowed scanning of patients weighing from 350 to 490 lbs. Engineering developments at that time focused on low and mid-field MR systems with an open configuration of the permanent magnet at field strengths up to 1.0T.47-49 MR systems with open designs can significantly reduce the anxiety of claustrophobic patients during MR scanning compared to conventional MR scanners.49 It is also technically much easier to perform MR-guided biopsies or ablations using an open MR system. Unfortunately, low and mid-field MR scanners remain restricted by their lower temporal and spatial resolutions (due to poor SNR). The functional imaging techniques available with such systems are also limited. Recently horizontal, wide bore high field (1.5T or 3.0T) MR systems with an open design have been developed. These offer the same benefits in patient comfort as the low field open systems of the past, while facilitating the implementation of high-quality functional MRI with diffusion (DWI) and perfusion-weighted imaging (PWI). 50-52
Receive Coil Technology
Receive coils are an essential component of the MRI hardware. They are usually comprised of one or more loops of conductive wire and are used to detect or receive MRI signals from the body. There are many different types of receive coils that can be employed for imaging, with specialization in part on the basis of body region. These include linearly polarized coils, circularly polarized coils, phased array coils, and matrix coils. Receive coils were long utilized in NMR prior to the advent of MRI. However, a 1976 paper by David Hoult presented a clear conceptual picture of the factors governing the SNR of an NMR experiment, thus beginning the modern era of NMR coil technology.53 The earliest receive coils employed for MRI were described by Ackerman et al in 1980.54 In that study, flat, round conductive loops (linearly polarized) were utilized for in vivo MR spectroscopy (MRS). This type of coil would eventually allow for the imaging of small superficial structures with improved SNR and spatial resolution.55,56 However, this type of surface coil was implemented almost exclusively as a receive-only coil. This was due to the reduction in transmit signal strength with increasing distance from the coil surface. Surface coils were also restricted to a relatively small imaging volume.
David Hoult explored the idea of a quadrature coil in 1984,57 a concept which he eventually patented. A quadrature coil is circularly polarized and consists of two linearly polarized coils utilizing two separate radiofrequency (RF) pulses. The pulses have a phase shift of 90o and are applied in orthogonally to each other in space. Circularly polarized coils can be used as a transmit and/or receive coil. A typical implementation of this coil configuration was the so-called birdcage coil that was commonly utilized for head imaging and MRS for many years.58-61
In 1990, Roemer et al described phased array coils, and demonstrated that greater SNR in spine MRI could be obtained with use of a four-element linear spine array.62 The basic design of a phased array coil combines multiple coil elements with multiple receive channels using circular polarization. The advantages of these coils include improved SNR and a greater potential imaging volume. In addition, multiple coil elements and multiple independent receive coils can acquire MR echoes simultaneously, and spatial information (phase-encoding [PE] steps) can be inferred from the geometric arrangement of the coils. Phased array coils (multi-element and multi-channel) thus eventually allowed for the implementation of parallel imaging (discussed in more detail below)-a technique providing substantial reductions in image acquisition times.63,64 The most recent developments in receive coil technology include phased array coils with an increasing number of coil elements for dedicated clinical applications. 32 to 64 channel systems are the current industry standard.65,66 Recent studies have utilized multi-element coils with 96, and even 128 channels, with several 128 channel 3T clinical systems currently installed in the United States.67-70
Since the earliest implementations of MRI, artifacts from patient motion have presented a hindrance to high-quality imaging. Navigator echo techniques were developed to reduce such artifacts, allowing for the monitoring and correcting of motion effects during MR acquisitions by tracking the movement of objects within the field of view (FOV) in real time. In initial studies by Ehman and Felmlee in 1989, "navigator" echoes were interleaved with imaging sequences to record positional displacement during the MR examination. These results demonstrated that navigator echoes could correct for gross object motion in both phantom and in vivo studies.71-73 Since then, this technique has been widely-utilized, in particular with respect to cardiac MRI. Navigator echo techniques allow for image acquisition during free breathing through the real-time monitoring of moving structures, typically the right hemidiaphragm.74-76 Image acquisition can then proceed based on direct measurements of the position of the diaphragm-lung or heart-lung interface, a technique advantageous relative to previous methods requiring a respiratory gating belt. Image quality, effective spatial resolution, and patient tolerance of the examination are substantially improved with the use of navigator echoes. In addition to motion correction, navigator echoes have been utilized to monitor kinematic abnormalities of the diaphragm, for example, in the setting of suspected diaphragmatic paralysis.77 Novel applications of navigator echo techniques with different image reconstruction algorithms continue to be developed.78-81
Physiologic motion during MR acquisitions results in a variety of artifacts. One method to resolve this problem is with the use of gating. The most common MR gating techniques in clinical use are electrocardiographic (ECG), finger plethysmographic, and respiratory gating. In 1984, Lanzer et al obtained high quality cardiac images using ECG and finger plethysmographic gating to acquire data during a single phase of the cardiac cycle.82,83 In that study, images acquired with plethysmographic gating were found to be similar in quality to those obtained by ECG gating. Subsequently, Runge et al described a respiratory gating technique.84,85 By acquiring data between the end of expiration and the beginning of inspiration, respiratory motion artifacts and motion blurring were significantly reduced. Simultaneous application of both ECG and respiratory gating allows for a detailed depiction of cardiac anatomy by MR. Additionally, reductions in artifacts from CSF pulsation have been observed with the use of peripheral gating techniques.86 The primary disadvantage of MR gating is increased scan time. Recently, investigators have focused on evaluating diffusion coefficients at different time points of the cardiac cycle using a temporally-resolved ECG-gated DWI sequence.87
Spin Echo Imaging
Spin echo (SE) imaging is a sequence that, in its most basic application, uses a 90Â° RF excitation pulse followed by one or more 180Â° refocusing pulses. SEs were initially introduced as an NMR sequence by Hahn in 1950.3 In that study, Hahn used three successive 90Â° pulses to produce a stimulated echo. In 1954, Carr and Purcell developed the first SE sequence consisting of two RF pulses, an initial 90Â° and subsequent 180Â° refocusing pulse.88 In 1958, the Carr-Purcell-Meiboom-Gill (CPMG) sequence was developed by Meiboom and Gill.89 This method incorporated multiple SEs for the first time and was implemented to measure T2 and diffusion. In 1981, Young et al first reported a clinical application of SE imaging for MRI of the posterior fossa.90 The SE technique applied in that study resulted in little gray-white matter contrast but excellent soft-tissue visualization at the skull base. In a subsequent study, Bailes et al utilized a SE sequence with a longer spin-spin relaxation time to detect brain abnormalities.91 That study demonstrated the sensitivity of SE imaging for the detection of a variety of neurological abnormalities such as lupus and parenchymal ischemia. By changing the repetition (TR) and echo times (TE) of SE sequences, image contrast is readily manipulated for various diagnostic purposes. 92,93 Today, variations of SE imaging techniques remain some of the most commonly utilized sequences in clinical MRI.
Fast/Turbo Spin Echo Imaging
Fast spin echo (FSE, or turbo spin echo [TSE]) was derived from the rapid acquisition relaxation enhanced (RARE) technique. RARE was initially proposed by Hennig et al in 1986 and further optimized by Mulkern et al in 1990.94,95 The basic pulse sequence design of RARE involves multiple 180o refocusing pulses that are applied after the 90o pulses in each TR interval. Flexible image contrasts can be achieved by reordering of the phase encoding steps. FSE allows much more efficient image acquisition than SE techniques, resulting in scan time reductions theoretically proportional to the echo train length (i.e. turbo factor).96 These scan time reductions are typically leveraged to obtain more signal averages, higher resolution images, or additional complementary sequences, all of which can result in improved image quality and diagnostic performance.96,97 Multiple 180o refocusing pulses render FSE sequences relatively insensitive to susceptibility artifacts. This may improve anatomic visualization in regions where there are pronounced differences in tissue susceptibility or in areas near metallic implants. The decreased effects of magnetic susceptibility on the final clinical image does result in a decreased sensitivity for the detection of blood products such as deoxyhemoglobin and hemosiderin.98 The high signal intensity of fat with FSE T2WI may also interfere with lesion visualization, and as such fat suppression (FS) techniques are commonly applied.97 FSE sequences rapidly-replaced SE PDWI and T2WI for imaging of the brain and spine.99 Meanwhile, a promising variation of TSE called half-Fourier single-shot turbo spin echo (HASTE) by Siemens was described by Semelka et al in 1996.100 HASTE is a type of ultrafast T2-weighted sequence (known as SS-FSE by General Electric [GE] and single shot TSE by Philips) that can be acquired in less than 1 second. Three-dimensional FSE techniques (discussed further below) used frequently in neurological and musculoskeletal applications, provide high SNR and allow image acquisition with isotropic voxel sizes permitting high-quality image reconstructions in arbitrary planes.101-103
Gradient Echo Imaging
Soon after the implementation of the "spin warp" methodology, gradient echo (GRE) techniques were developed, appearing in the MRI literature as early as 1981.90 These sequences enabled, for the first time, visualization of arteries and veins in the posterior fossa without the use of contrast agents. However, initial GRE sequences were performed using 90Â° flip angles (FA) and TRs comparable to SE imaging techniques, on the order of hundreds of milliseconds. Rapidly-acquired GRE sequences employing short TRs and FAs less than 900 were developed in the mid-1980s, and Jens Frahm patented the technique in 1985. These sequences rapidly generate MR signals by using a pair of bipolar gradientÂ pulses. Spoiled and refocused GRE sequences were most commonly utilized; spoiled GRE imaging was introduced by Haase, Frahm, and Matthaei in 1986.104,105 This group described a fast low-angle shot (FLASH) sequence (the GRE induced by reversal of the readout gradient) which, due to its rapid acquisition, allowed acquisition of abdominal MRI without significant artifacts from respiratory motion. Subsequently, Evans investigated pulsatile flow using a gradient refocused acquisition in steady state (GRASS) sequence.106 In cardiovascular imaging, the rapid acquisition times with GRE sequences facilitate imaging at multiple time points in the cardiac cycle.107 As GRE sequences are sensitive to field inhomogeneities resulting from magnetic susceptibility effects, they can used to detect acute or chronic intracranial hemorrhage not evident with SE imaging.108 A recent study demonstrated acquisition of high quality abdominal MRI with a free-breathing radial 3D FS T1-weighted gradient-echo (radial VIBE) sequence, a potentially beneficial technique for imaging of patients who cannot comply with breath-hold instructions.109
2D Multi-slice Imaging
2D multi-slice imaging is a technique whereby each repetition period is utilized for the simultaneous acquisition of multiple slices. This results in a decreased acquisition time for multi-slice MR acquisitions. 2D multi-slice imaging was introduced by Crooks et al in 1982 on a 0.35T MR system.110 Using this technique, the authors demonstrated acquisition of multiple sequential imaging slices in a timeframe typically needed for acquisition of a single slice. Subsequently, multi-slice acquisitions were utilized for evaluating the cervical spine, and were found to be particularly helpful in the detection and characterization of extradural lesions.111,112 As inappropriate slice selection can degrade image contrast with T2-weighted techniques, the choice of inter-slice gap was found to be crucial113,.114 2D multi-slice imaging has proven to be a versatile technique, adaptable for use in combination with GRE, FSE (TSE), inversion-recovery (IR), and steady-state free precession (SSFP) imaging.112,115,116 The sensitivity (i.e. SNR per unit of time) of 2D multi-slice techniques is similar to that of 3D imaging sequences.117 Today, 2D multi-slice techniques are widely used, with continuing development in specialty applications, for example in cardiovascular MRI when rapid image acquisitions are required.118
Three-dimensional MR imaging was introduced by Kramer et al in 1981.119 However, its initial clinical application was restricted due to long scan times compared to conventional 2D MR imaging. In 1986, Frahm et al designed a sequence enabling rapid 3D MR imaging. This technique was based on the FLASH sequence, which at the time had already been successfully utilized in neurological and musculoskeletal imaging.120 3D FLASH was a GRE sequence with a relatively rapid acquisition time due to short TE and TR. This 3D imaging technique allowed isotropic image acquisition with thin continuous slices (without interslice gaps). Multi-planar images could thus be constructed from a single acquisition with only minimal partial volume effects.121,122 In 2001, SSFP-another GRE-based 3D imaging technique-was introduced for cardiac imaging.123 This technique combined with various methods of fat saturation can provide T2/T1-weighted contrast. Because of the high signal intensity of intravascular blood on this sequence and its fast acquisition time, SSFP has become the technique of choice for acquisition of cine cardiac MRI. 3D imaging techniques based on GRE sequences are currently used as the primary sequences for MRA and breath-hold thoracoabdominal imaging.109,124,125 3D FSE sequences have also been implemented for brain and musculoskeletal imaging.101,126 Improved image contrast and decreased susceptibility artifacts with 3D FSE techniques can be advantageous relative to 3D GRE techniques.127,128
Partial Fourier or Half Fourier MR Imaging
Partial Fourier MR imaging was introduced by Feinberg in 1986.129 This technique reduces acquisition time by acquiring only a fraction of k-space data, utilizing a mathematical reconstruction to fill the remainder. The algorithm is based on the inherent symmetry of k-space. One potential technical drawback to this technique is that spatially dependent phase shifts can cause data to be asymmetrical, producing artifact. Available implementations of partial Fourier imaging can correct these phase shifts and reduce associated artifacts by acquiring supplemental data.130
Partial Fourier MR imaging can reduce scan time by up to half while still preserving spatial resolution. The primary disadvantage with this approach is a reduction in SNR. In 1990, Runge and Wood reported that the SNR of white matter, the CNR between gray and white matter, and the ability to detect lesions in multiple sclerosis patients with partial Fourier MR imaging was diminished compared to a conventional acquisition using a full complement of phase-encoding steps.131 Partial Fourier acquisition techniques nevertheless remain promising and are commonly implemented in clinical practice. Recent investigations have shown that HASTE DWI can achieve greater contrast between the central and peripheral zones of the prostate, more accurate apparent diffusion coefficient (ADC) values, and greater reproducibility compared to echo planar imaging (EPI).132
Inversion recovery sequences were originally employed for T1-weighted imaging. In 1981, Doyle et al and Young et al utilized IR sequences to obtain heavily T1-weighted images in clinical patients.133 134 These studies demonstrated that IR sequences could successfully depict many brain lesions not evident by computed tomography. Ehman et al used an IR sequence to assess T2 time dependence in 1984.135 That study showed that T2 dependence might enhance or diminish the contrast between two tissues in an IR-based image. Increased scan time was an initial detriment of IR sequences; however, as faster pulse sequences were developed, IR techniques were combined with FSE, rapid GRE, and EPI to reduce acquisition times into more acceptable ranges.136-138
Many variations on IR can be utilized for clinical imaging. A short-tau (i.e. time of inversion [TI] of approximately 125-250 ms) inversion recovery sequence (STIR) can be used for fat-suppressed body and musculoskeletal imaging, providing a high degree of soft-tissue contrast. Medium range TI (approximately 250-700 ms) sequences are employed in brain and spine imaging to improve brain-cerebrospinal fluid (CSF) contrast. Selection of a medium range TI also allows acquisition of high-contrast hepatic imaging. IR sequences with a long TI (up to 700 ms and longer), i.e. fluid attenuated inversion recovery (FLAIR), are useful in brain MRI, as the suppression of CSF signal allows for improved lesion detection. IR sequences are also utilized today to image blood products and to measure tissue perfusion.139,140
Fluid attenuated inversion recovery is an IR pulse sequence that reduces signal from fluids with very long T1 and T2 times like CSF. This technique was proposed by Bydder et al in a 1982 paper.141 By suppressing CSF signal, FLAIR effectively highlights periventricular and subcortical lesions.139,142,143 Most clinical applications are combined with heavily T2-weighted imaging utilizing a long TE readout to improve the conspicuity of brain parenchymal abnormalities. FLAIR techniques can also be implemented with T1-weighted imaging. T1 FLAIR improves gray-to-white matter contrast, providing higher spatial resolution with thin sections. Adequate contrast for routine brain or spine imaging can be obtained with T1 FLAIR with shorter scan times than conventional 2D SE sequences.144 Despite these advances, FLAIR has historically been restricted by long acquisition times, thus necessitating the use of fast imaging techniques such as FSE or an interleaved IR sequence to acquire two-dimensional images.145,146 With the recent development of variable flip angle techniques, 3D FLAIR has also proved to be valuable in brain imaging.147-149 3D FLAIR reduces motion artifacts and optimizes lesion detection by using contiguous thin slices and high SNR multi-planar image reformats. At field strengths of 3T and greater, the prolonged T1 relaxation times of brain parenchyma and relatively constant T1 of CSF result in diminished contrast between the two tissues. As a result, T1 FLAIR is often the sequence of choice for imaging of the brain and spine and 3 T. Applications with contrast-enhanced FLAIR have also proven to be useful for improved detection of brain parenchymal tumors and meningeal lesions.150
Periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) MRI is a technique for motion correction that functions by acquiring data in concentric rectangular strips that rotate around the center of k-space. This technique is also referred to as BLADE on Siemens Healthcare based MRI systems. PROPELLER MRI was first described by James Pipe in 1999.151 In that work, Dr. Pipe found that PROPELLER MRI reduced both head and respiratory motion artifacts. Later studies found that pulsation, ghosting, and susceptibility artifacts are also reduced with PROPELLER MRI.152-154 Naganawa likewise showed that flow-related artifacts were greater in T1 SE imaging than that with T1 FLAIR BLADE.153 Wintersperger et al also found substantial improvements in artifacts with T2-FLAIR BLADE compared with standard FLAIR.152
Oversampling the central area of k-space results in an improvement in SNR with the PROPELLER technique, but at a cost of longer scan and image reconstruction times compared to conventional sequences. Nevertheless, the PROPELLER sequence has proven successful clinically. Forbes et al found multi-shot FSE PROPELLER DWI to improve image quality and visualization of acute cerebral infarctions relative to EPI DWI.155 Hirokawa et al showed that the implementation of PROPELLER with fat-saturated T2-weighted imaging obtained after injection of super paramagnetic iron oxide offered improved detection of hepatic metastases.156 PROPELLER MRI has also proven useful for quantifying and compensating for head motion during MRI acquisitions.157
Magnetization prepared rapid gradient echo (MP-RAGE) is an ultrafast GRE sequence with an incorporated magnetization preparation pulse, described by John Mugler in 1990.158 A variety of trademarked names now refer to a similar technique (MP-RAGE by Siemens, 3D FGRE by GE, and 3D-TFE by Philips). In Mugler's work, MP-RAGE was utilized to generate high-quality 3D images of the head and abdomen. Through manipulation of magnetization preparation pulses, the sequence can provide either heavily T1 or T2-weighed image contrast.159 Compared to conventional SE sequences, Mugler et al found that 3D MP-RAGE increased cerebral gray/white matter signal and contrast-to-noise (CNR) by more than 50% on heavily T1-weighted images.160 De Lange et al subsequently implemented 2D MP-RAGE for MR imaging of focal liver lesions. MP-RAGE demonstrated CNR comparable to that of conventional SE imaging, but with fewer artifacts from respiratory motion.161 Meanwhile, Runge et al examined 3D MP-RAGE relative to FLASH techniques for brain imaging. In that work, similar T1 contrast and CNR were observed between the two sequences; however, imaging time and motion artifacts were reduced with MP-RAGE.162 MP-RAGE acquisition times may still be too long for successful abdominal imaging in non-cooperative patients, particularly when FS is required. For such cases, Altun et al proposed an MP-RAGE technique for water excitation post-contrast imaging.163 Another important variant of MP-RAGE is termed magnetization-prepared spiral acquisition gradient-echo (MP-SAGE). This technique utilizes interleaved square-spiral phase encoding to further increase SNR and CNR for brain imaging.164 More recently, an optimization of a two inversion-contrast magnetization-prepared rapid gradient echo sequence (MP2RAGE) has been introduced in an evaluation of early-stage multiple sclerosis.165
FSE Imaging with Flip Angle Evolution
More recent FSE and TSE sequences allow for high-resolution imaging with relatively short acquisition times and low specific absorption rates through use of long echo trains and variable FAs. The major MR vendors employ various trademarked names for this type of sequence, known as SPACE (Sampling Perfection with Application optimized Contrasts using different flip angle Evolutions) by Siemens, CUBE by GE, and VISTA (Volumetric Isotropic TSE Acquisition) by Philips. SPACE was initially described by Mugler et al for neuroimaging in a 2000 article in Radiology.166 In 2005, Lichy et al successfully applied SPACE outside the brain.167 That study showed that SPACE could provide high spatial resolution, isotropic 3D T2-weighted TSE MR imaging in the pelvis, spine, and extremities. In a subsequent study, Arizono et al used SPACE for 3D MR cholangiography.168 The study showed that high-resolution 3D MR cholangiography with SPACE at 3T allowed for high-quality imaging of the biliary tract in healthy volunteers. The advantages of SPACE and similar sequences include: (1) high SNR, (2) improved spatial resolution, (3) relatively short acquisition times, (4) low specific absorption rate, (5) flexible image contrast, and (6) the ability for multiplanar reconstructions.101,167,169-174 SPACE has been widely used for imaging of the entire body, including head, spine, abdominal, pelvic, orthopedic, and vascular imaging.167,168,170,172,174-177 The latest modification of this technique involves the use of multiple slab acquisitions with view angle tilting based (MSVAT-SPACE) to correct for metallic artifacts.18
Echo planar imaging is a pulse sequence that provides ultrafast MR imaging. EPI was developed by Mansfield et al in 1981 and was first utilized to obtain cardiac images in rabbits.178,179 Shortly after its development, EPI became widely used for advanced MRI applications such as DWI, PWI, and fMRI, techniques initially applied in the brain and cardiovascular system.180,181
Image acquisition with EPI is more rapid than with other pulse sequences: all or a large number of PE lines of k-space are acquired from a single RF excitation as opposed to the typical acquisition of one PE line per each RF excitation. When combined with SE, GRE, or IR sequences, EPI offers greater temporal resolution and less motion sensitivity compared to conventional MR techniques. This allows for in vivo imaging of rapid physical processes.
EPI can be performed using a single excitation (single-shot) or multiple excitations (multi-shot). Single-shot EPI acquires data more rapidly, and with less motion artifact. However, off-resonance effects can reduce image quality secondary to susceptibility and chemical shift artifacts, a shortcoming that has been subsequently reduced by the development of high-performance imaging gradients. In 1994, multi-shot EPI was developed based on an interleaved EPI method.182 Multi-shot EPI techniques use multiple RF excitations to fill k-space, improving SNR and reducing geometric distortion. However, multi-shot techniques require longer acquisition times and can be limited by artifacts from patient motion. However, readout-segmented EPI methods and 2D navigator phase correction techniques have been recently applied to multi-shot EPI minimizing motion artifacts and improving spatial resolution.
Time of flight magnetic resonance angiography (TOF MRA) was initially introduced by Laub et al in 1988.183 In this method, the signal of stationary tissue in the volume of interest is saturated by the use of a very short TR, whereas unsaturated inflow spins are detected as high signal intensity using a flow-compensated GRE sequence. This technique provides a noninvasive method for imaging the cardiovascular system and has attracted the attention of investigators and clinicians since its introduction. Initially, most research focused on evaluating diseases of the carotid and cerebral vasculature.184-186 Results were encouraging; although, 3D TOF MRA was initially performed using a single large imaging volume resulting in relatively poor spatial resolution. Over time, multi-slab 3D TOF MRA became possible, improving spatial resolution and decreasing flow-rate dependence.184,187 With today's techniques, 3D TOF MRA is able to detect the nidus of a cerebral arteriovenous malformation with diagnostic quality approaching that of a conventional angiogram. 3D TOF MRA can accurately detect arterial stenoses; although, some flow-related limitations remain.184,185 Meanwhile, 2D TOF magnetic resonance venography (MRV) allows imaging of the cerebral veins and venous sinuses, detecting intracranial venous thrombosis with diagnostic accuracy similar to DSA.186
Contrast-enhanced MR Angiography
Contrast-enhanced MR angiography (CE MRA) is a technique that exploits ultrafast MR imaging methods, achieving flow-independent vascular imaging via intravenous injection of a gadolinium chelate (Gd) contrast agent. CE MRA was demonstrated by Revel et al in 1993.188 In this landmark study, the authors described a novel imaging technique using a 2D ultrafast GRE sequence following rapid intravenous bolus injection of gadoterate meglumine (Dotarem). In 1994, Loubeyre et al successfully applied 2D turbo FLASH CE MRA to pulmonary arterial imaging for the evaluation of pulmonary embolus.189 These studies revealed the utility of CE MRA in assessing the great vessels. With progress in MR techniques, 3D GRE sequences eventually replaced 2D turbo FLASH. Prince et al introduced 3D CE MRA in 1994,190 using a 3D Fourier-transform spoiled gradient-echo volume sequence with a bolus injection of gadolinium chelate. Hardware improvements have allowed for further reductions in the TE utilized for CE MRA, while parallel imaging techniques enable reductions in imaging times and acquisition of 4D images.191,192,193 Today, CE MRA is commonly implemented to assess both the arterial and venous systems in many anatomic regions including the carotid, coronary, and renal arteries as well as the aorta and peripheral vasculature.125,194-197
Phase-contrast (PC) imaging is an MR technique that is sensitive to flow-induced phase shifts and allows for the quantitative measurement of flow within arteries, veins, and CSF. Initially described by Moran in 1982 as a "bipolar velocity-encoding gradient pulse,"198 PC techniques incorporate phase-modulation into a conventional MR acquisition. This modulation suppresses phase accumulation for stationary tissues but facilitates phase accumulation in tissues with moving spins (i.e. flowing blood). Two data sets with and without flow-encoding are acquired and then subtracted to obtain a map of flow velocity (i.e. PC images).199,200 Typically, a PC sequence yields two types of image information: a phase image, in which the signal intensity (bright or dark) is related to the both flow velocity and direction, and a magnitude image, in which the signal intensity is only related to the flow velocity. PC techniques are mainly used for MR angiography (PC MRA),201 acquired either by multi-slice 2D or volumetric 3D imaging techniques.202,203 Compared to 2D acquisitions, 3D PC MRA can reduce scan times and eliminate misregistration artifacts due to patient motion.204 PC MRA also has the advantages of superior background suppression and improved depiction of areas with slow intravascular flow, providing more direct flow information than TOF MRA.205 With the development of rapid MR acquisition techniques, time-resolved PC MRA (4D PC MRA) has recently been developed.206,207 This new method allows depiction of complex flow patterns within the cardiovascular system over time.208,209
Dynamic CE MRA
Dynamic CE MRA was introduced by Revel et al in 1993.188 The method in that article combined sub-second MR imaging with a rapid bolus injection of an MR contrast agent to acquire 10-20 images in a single plane during a single breath hold. Thus, not only vascular morphology was depicted, but also dynamic flow-related information. During this time period of development, dynamic information over a number of slices was only possible to be acquired using 2D Turbo-FLASH.188,189 The advent of 3D time-resolved imaging of contrast kinetics (TRICKS) was a further breakthrough in contrast-enhanced MRA. With this method, multiple 3D image sets could be acquired during a single breath-hold, leading to improved arterial or venous imaging that is less dependent on bolus timing.210 Combining parallel imaging with a k-space undersampling strategy further improved the temporal resolution of 3D dynamic CE MRA.211 Image subtraction could be performed with 3D TRICKS to allow for pure arterial phase imaging.210 TRICKS is the vendor-specific term for the GE version of this type of sequence which is also known as TWIST (Time-resolved angiography With Interleaved Stochastic Trajectories) in its implementation by Siemens. Recent research has shown 3D dynamic CE MRA to be helpful in delineating the feeding arteries of tumors and in depicting intrinsic tumor vascularity in the head and neck.212
Several techniques for fat suppression exist.213 The most common include: (1) Dixon, (2) frequency-selective fat saturation (chemical shift selective saturation, CHESS), (3) STIR, and (4) water excitation (spatial-spectral pulse).214-217
In 1984, Dixon reported that simple spectroscopic imaging techniques could separate water from fat.214 The Dixon technique relies on water/fat chemical shift differences to generate in and out of phase images. FS water only images could thus be acquired by a simple summation of the two images. Many obstacles hindered the earliest use of this technique, including sensitivity to B0 inhomogeneities, longer acquisition times, increased motion artifacts, and image blurring.218 Since then, the Dixon technique and its many variations have been further developed. In modern imaging the Dixon technique has been used in liver acceleration volume acquisition (LAVA, also known as VIBE and THRIVE) to generate dynamic CE MRI of the abdomen.219
In 1985, Haase et al introduced the CHESS technique as a method to selectively suppress unwanted tissue signal components.215 Using this technique, signal from a given type of tissue can be suppressed by applying a frequency-selective 90° excitation pulse followed by a dephasing gradient. Haase et al demonstrated the ability to acquire fat-suppressed images of the human hand with suppression of bone marrow signal. Subsequently, Frahm et al and Matthaei et al applied the CHESS technique clinically.220,221 Frahm et al successfully acquired CHESS MRI of the healthy human head and hip, and Matthaei et al reported the use of CHESS for the assessment of femoral head avascular necrosis. Variations of the CHESS technique have since been widely used in imaging of the knee, pelvis, and abdomen.222 The primary disadvantage of CHESS is its sensitivity to B0 inhomogeneities.
In 1985, Bydder et al introduced use of the STIR sequence for FS.216,223 STIR was shown to provide high soft-tissue contrast for abdominal imaging through the suppression of fat signal. Compared with the Dixon and CHESS techniques, STIR has the advantage of producing uniform FS with relative strong insensitivity to B0 inhomogeneity. The disadvantages of STIR include reduced SNR and longer acquisition times. Additionally, STIR suppresses all signals resulting from short T1 times, and thus cannot be utilized for CE MRI.222
A more advanced technique for FS was introduced by Meyer et al in 1990.217 In that work, FS was performed by selectively exciting only water molecules. Water excitation can be used in conjunction with a variety of other sequences, including SE and FSE imaging as well as EPI, rapid GRE, and balanced SSFP imaging.163,224-226 Water excitation techniques are also relatively insensitive to B0 inhomogeneities, rendering this approach of particular value at higher field strengths. Water excitation techniques do require relatively lengthy acquisition times.222
Perfusion-weighted imaging is an MR technique that allows for non-invasive, high-resolution measurements of tissue perfusion at the microvascular or capillary level. In 1990, Belliveau et al demonstrated the ability to visualize cerebral perfusion by using the susceptibility effects of contrast agents in combination with ultrafast MRI techniques.227 Subsequently, Rosen et al presented a human study demonstrating cerebral blood volume (CBV) changes resulting from task activation and in the presence of brain tumors.228 Runge et al applied PWI to assessments of brain ischemia in 1993.229 The PWI technique described in these papers is now widely used in clinical practice and is reliant on the use of a Gd contrast agent in combination with rapid image acquisition. In this approach, susceptibility effects of the highly-concentrated contrast agent induce local T2* signal changes, causing a loss of signal intensity within tissue. Functional parameters measuring tissue perfusion, such as CBV, can then be mathematically calculated from a signal-time curve.230 Arterial spin labeling (ASL) is a second major approach for PWI as reported by Detre et al in 1992 using a rat model at 4.7T.231 With ASL, water molecules are labeled prior to arrival within the tissue of interest. This technique thus images T1 signal changes in magnetically labeled blood, which are directly related to absolute hemodynamic parameters (i.e. CBV and cerebral blood flow [CBF]). PWI currently represents a basic scientific and clinical tool in brain imaging.232 PWI can also be implemented based on measurements of incoherent intra-voxel proton motion.15,233,234
Diffusion-weighted imaging is an MRI technique that maps molecular diffusion coefficients. The NMR basis for DWI was originally conceived of by Carr and Purcell who observed the sensitization of NMR echoes to the effects of diffusion.235 In 1965, Stejskal and Tanner proposed a sequence for obtaining diffusion sensitization through application of short duration gradient pulses,236 forming the basis for image acquisition strategies used in clinical DWI MR today.237 The use of DWI for medical imaging was first reported by Wesbey et al in 1984.13,14 Initially, "molecular self-diffusion coefficients" were measured in vitro by varying slice-selective gradients during data acquisitions using an MR scanner. Two years later, Le Bihan performed DWI scans on healthy and diseased brains.15 Le Bihan first introduced the notion of ADC values which was deemed the superior descriptor because the diffusion measurements performed were influenced not only by diffusion changes on the molecular level but also by the microcirculation in the capillary network and fluid flow. Over the next several years, researchers focused on the DWI of neurological disorders with a great deal of success coming in the evaluation of acute stroke.238,239 Ischemic regions could be detected as early as 15 minutes following arterial occlusion using a DWI sequence with application of a very strong diffusion gradient (reflected by the b-value). At that time, ADC values were obtained by imaging with simply two different b-values. More recently, multi b-value DWI sequences have been described, providing greater accuracy in measuring tissue diffusion, with utility in particular in body imaging.240 Today, a great deal of DWI research is focused on the evaluation of abdominal pathology.241,242 Current research demonstrates that DWI improves tumor detection and can serve as a noninvasive biomarker of tumor aggressiveness.
Diffusion Tensor Imaging
Diffusion tensor imaging is an MR technique that measures and maps the orientation of myelin fibers based on their anisotropy. This technique was described in the early 1990s. Moseley successfully depicted the orientation of feline white matter by altering the directions of diffusion gradients in 1990.16 One year later, Douek obtained a color map of myelin fiber orientation in human volunteers, which was generated by combining two independent DWIs obtained with different diffusion gradient orientations.243 DTI images were finally acquired utilizing only a single scan by Mori in 1995.244 DTI allows quantitative evaluations of diffusion anisotropy and tracing of the diffusion tensor. From this, maps of the brain (white matter tractography) and other soft tissue structures can be constructed depicting ADC, relative anisotropy, and fractional anisotropy.245 Early research efforts with DTI found that alterations in ADC and diffusion tensor anisotropy reflected histopathological changes in the lesions of multiple sclerosis.245 Recent studies have shown that DTI can provide helpful diagnostic information in patients with prostate and breast cancer.246,247
Initial abdominal MRI techniques could not achieve image quality comparable to that obtained in the central nervous system. This was due, in part, to the effects of respiratory motion.248,249 Artifacts from respiration consist principally of image ghosting propagated along the phase-encoding direction with regular periodicity. Various methods have been proposed to compensate for such artifacts during abdominal MR imaging.
In cooperative patients, the simplest method is to utilize breath holding to eliminate respiratory motion. When patients cannot adhere to such instructions, other strategies are needed. Respiratory gating was initially evaluated in abdominal MRI by Runge et al and Ehman et al in 1984.85 84 These studies showed respiratory gating could eliminate gross motion artifacts, improving contrast and spatial resolution of abdominal MRI. Unfortunately, respiratory gating can increase the acquisition time by a factor of two or more, limiting its use. In 1985, Bailes et al introduced respiratory ordered phase encoding (ROPE) as a method for reducing respiratory ghosting from view-to-view motion in MR imaging.250 Lewis et al proposed respiratory triggering as an alternative to gating in 1986, concluding that triggering was less time consuming than gating and could be easily applied to sequences with short TR.251 Other methods, including gradient moment nulling (GMN), navigator echoes, and parallel imaging have also been used to minimize or eliminate artifacts from respiratory motion.252-254
The first MR image of the human heart was obtained by Hawkes in 1981.255 Since that time, cardiac MRI has become a part of routine clinical imaging. The technique provides a valuable assessment of cardiac morphology, function, myocardial viability, and even coronary anatomy. However, periodic cardiac motion and respiration, which may cause image blurring, ghosting and misregistration, remain a challenge for cardiac MR examinations. Various strategies have been proposed to compensate for cardiac motion. Gated cardiac MR imaging was initially evaluated by Lanzer in 1984.83 That study demonstrated ECG-gating to be a reliable technique for triggering the data acquisition relative to the use of other external sensors.256 By synchronizing image acquisition with the R-wave of the ECG, gated data could be prospectively acquired or retrospectively reordered for a given phase of cardiac cycle.257 In addition, other investigators reported that similar triggering information could be obtained from MRI data for motion correction. This "self-gated" technique is typically restricted to cardiac cine imaging because it requires continuous scanning to monitor cardiac motion.258-260 Meanwhile, respiratory gating,85 navigator echoes,73 and self-gating techniques260 have also been proposed to compensate for respiratory motion during cardiovascular or thoraco-abdominal imaging. Other strategies, including optimal k-space sampling261 and fast acquisition sequences with parallel imaging,262 have also been proposed to diminish the effects of motion. Today, ECG-gating remains a commonly used technique for cardiac imaging, as do other methods of triggering the MR acquisition to coincide with portions of the cardiac cycle.87,263
The presence of paramagnetic deoxyhemoglobin in venous blood generates variations in magnetic susceptibility within the vessel and its surroundings. As a natural occurring contrast, deoxyhemoglobin can thus be used to assess the function of tissue by noninvasively monitoring real-time blood oxygenation levels in vivo.17 In 1990, Ogawa acquired MR images using BOLD contrast in mice and rat brain studies by employing a GRE sequence at high field strengths (7T and 8.4T).264 Two years after the introduction of BOLD, Kwong utilized the technique to obtain maps of human brain activity based on signal changes during resting and stimulated sensory states.265 At that time, most researchers were interested in detecting cortical activation in the visual association areas during different task stimulations.266,267 Recently, investigators have used BOLD MRI to study renal function.268,269 With administration of diuretics, BOLD MRI has been shown to distinguish oxygenation levels in the renal cortex and medulla. Additionally, some investigators have applied BOLD MRI to evaluate the function of human skeletal muscle in response to ischemia/reactive hyperemia.270
Magnetic resonance proton spectroscopy is the most commonly utilized MR-based method to noninvasively evaluate metabolic changes in the human body. High-resolution in vivo 1H MRS