Show Summary Details
Page of

Common MR Imaging Artifacts 

Common MR Imaging Artifacts
Chapter:
Common MR Imaging Artifacts
Author(s):

Kiaran P. McGee

and Matthew A. Bernstein

DOI:
10.1093/med/9780199941186.003.0012
Page of

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). © Oxford University Press, 2016. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 17 December 2018

Cardiac magnetic resonance (MR) imaging involves the use of a complex and sensitive imaging device to obtain high temporal- and spatial-resolution images of the heart as it changes shape, size, and position throughout the cardiac cycle. Because of this inherent complexity, measurement errors inevitably occur, which results in artifacts within the reconstructed image. Even when the MR scanner is functioning correctly, inappropriate choice of specific imaging parameters also can result in image artifacts. Successful cardiac imaging, therefore, requires knowledge of not only the most commonly encountered artifacts and their sources but also methods to reduce them. The purpose of this chapter is to describe several commonly encountered imaging artifacts, their causes, and countermeasures to either reduce or ameliorate them.

Aliasing

Source

Aliasing occurs when the field of view (FOV) in the phase-encoding direction is smaller than the anatomy being imaged. The data are undersampled and the image appears to be wrapped from one side to the other along this direction (Figure 12.1). This can occur along 2 directions when 3-dimensional data are acquired, but it does not occur along the frequency-encoding direction.


Figure 12.1 Aliasing. A, Reduced-phase FOV 4-chamber image of the heart demonstrating aliasing of the chest wall and arms into the anatomical region of interest. B, Full-phase FOV image of the heart with phase- and frequency-encoding directions swapped.

Figure 12.1 Aliasing. A, Reduced-phase FOV 4-chamber image of the heart demonstrating aliasing of the chest wall and arms into the anatomical region of interest. B, Full-phase FOV image of the heart with phase- and frequency-encoding directions swapped.

Solutions

  • Increase the FOV along the phase-encoding axis. This will solve this problem but will decrease the spatial resolution along this direction.

  • Enable no-phase-wrap or phase-oversampling features.

  • Swap the readout and phase-encoding directions.

  • Apply parallel imaging techniques to “unwrap” the artifact. This will solve the problem but can decrease signal-to-noise ratio in the image.

  • Reposition the center of the FOV so that the aliased regions are not included in the anatomy of interest. Signal-to-noise ratio and resolution are not decreased, but several attempts may be required to position the aliased tissue outside the region of interest.

Gradient Decay or Fall-Off

Source

Gradient decay or fall-off is a result of the limited physical extent of the spatial-encoding gradient fields, most notably along the physical Z-axis of the MR scanner, which, in a cylindrical-bore magnet, corresponds to the head-foot direction of the patient. Along this axis, the gradient field is at a maximum at the physical ends of the coil but decreases to zero along this axis away from the ends of the coil. Radiofrequency (RF) excitation and reception of signal from tissue within the fall-off zone of these gradient fields results in tissue within this region being spatially mismapped on the reconstructed image (Figure 12.2). This effect is due to aliasing, so it occurs along the phase-encoding direction and can appear as different types of artifacts. Several terms, including peripheral signal artifact, feather artifact, and cusp artifact (also known as “annefact”) have been used.


Figure 12.2 Gradient decay or fall-off. A and B, Single-shot, fast spin echo, sagittal, 35-cm FOV localizer image with a superior-inferior (A) and anterior-posterior (B) phase-encoding direction. Image A shows aliasing (wrap) of anatomy within the decay zone of the gradient and included within the RF torso coil, with gradient fall-off contributing to the spatial distortion and hyperintensity in regions of the artifact. Switching phase and frequency axes effectively eliminates the artifact in this instance. C, Sagittal spin-echo, T2-weighted spine image shows “feathering” (arrow) along the superior-inferior direction at the position of the vertebral bodies, indicating the presence of annefact (cusp) artifact. An element of the multielement RF coil is located within the fall-off region of the Z-axis gradient and is incorrectly spatially encoded, thereby producing the feathering along the phase-encoding direction. D, Short-axis, balanced steady-state free precession (SSFP) image of the heart demonstrating both feathering and wrap from anatomy outside the imaging plane. Although annefact artifact (arrows) is most commonly associated with multiecho spin echo sequences, this image demonstrates that other pulse sequences can reproduce this effect. Due to fall-off of the gradients, fat signal in the aliased anatomy (arrowhead) is brighter than other fat and is due to the collapse of this signal from anatomical regions that would normally be spatially encoded.

Figure 12.2 Gradient decay or fall-off. A and B, Single-shot, fast spin echo, sagittal, 35-cm FOV localizer image with a superior-inferior (A) and anterior-posterior (B) phase-encoding direction. Image A shows aliasing (wrap) of anatomy within the decay zone of the gradient and included within the RF torso coil, with gradient fall-off contributing to the spatial distortion and hyperintensity in regions of the artifact. Switching phase and frequency axes effectively eliminates the artifact in this instance. C, Sagittal spin-echo, T2-weighted spine image shows “feathering” (arrow) along the superior-inferior direction at the position of the vertebral bodies, indicating the presence of annefact (cusp) artifact. An element of the multielement RF coil is located within the fall-off region of the Z-axis gradient and is incorrectly spatially encoded, thereby producing the feathering along the phase-encoding direction. D, Short-axis, balanced steady-state free precession (SSFP) image of the heart demonstrating both feathering and wrap from anatomy outside the imaging plane. Although annefact artifact (arrows) is most commonly associated with multiecho spin echo sequences, this image demonstrates that other pulse sequences can reproduce this effect. Due to fall-off of the gradients, fat signal in the aliased anatomy (arrowhead) is brighter than other fat and is due to the collapse of this signal from anatomical regions that would normally be spatially encoded.

Solutions

  • For the coronal and sagittal imaging planes, select the frequency-encoding direction to be along the superior-inferior direction. If this causes flow and breathing artifacts to propagate over the anatomy of interest, then one of the other countermeasures should be used.

  • Use receive-only coils that do not extend beyond the anatomical region of interest.

  • If anatomical coverage is large and multicoil element surface coils are used, select only the coils that cover the region of interest.

  • Place saturation bands over the regions in the fall-off zone of the gradient fields.

  • Use transmit-receive instead of receive-only coils. For cardiac applications, this is usually not practical because the body RF coil typically must be used for transmission.

RF Zipper

Source

An RF zipper results when RF energy from a source other than the patient is detected by the RF coil used for imaging. The signal is not spatially encoded and appears as a line whose intensity varies from bright to dark along the phase-encoding direction (Figure 12.3). The typical source of this noise is (but is not restricted to) a pump or other electromechanical device in the room. Also, any conducting cable that enters the room can act as an antenna, propagating noise from outside into the scan room. Finally, a leak in the RF shield of the MR scan room can allow external RF noise to enter. These RF leaks occur most commonly around doors and windows.


Figure 12.3 RF Zipper. Four-chamber, balanced SSFP, gradient echo (A), perfusion (B), and late gadolinium-enhancement (C) images showing an RF zipper propagated along the phase-encoding direction. The noise source that produced this artifact was an infusion pump in the MR scan room. For imaging sequences with low signal-to-noise ratio, such as late gadolinium-enhancement imaging, the artifact can be particularly prominent.

Figure 12.3 RF Zipper. Four-chamber, balanced SSFP, gradient echo (A), perfusion (B), and late gadolinium-enhancement (C) images showing an RF zipper propagated along the phase-encoding direction. The noise source that produced this artifact was an infusion pump in the MR scan room. For imaging sequences with low signal-to-noise ratio, such as late gadolinium-enhancement imaging, the artifact can be particularly prominent.

Note that zipper-type artifacts that occur along the frequency-encoding axis arise from sources internal to the MR scanner such as stimulated echoes and pulse sequence timing errors.

Solutions

  • Identify and remove all noise sources within the room.

  • Have a qualified service engineer check for leaks in the RF shield of the scan room.

ECG Gating Mistriggers

Source

A poor quality (low-voltage) electrocardiographic (ECG) waveform or an irregular heart rate can degrade cardiac MR image quality. When the voltage of the QRS complex is nearly the same value as that of the noise, triggering on noise spikes instead of the R-wave peak causes data to be collected at incorrect phases of the cardiac cycle. The random nature of the noise results in mixing of data from multiple cardiac phases into a single phase and produces motion blurring, ghosting, and signal loss. ECG-gated MR sequences typically use some type of arrhythmia rejection to ensure that all data are collected at the correct cardiac phase. Arrhythmia rejection prolongs the data acquisition period, which results in longer breath-holds for the patient. If arrhythmia-induced heart rate changes are too great, the scanner may time out, resulting in no data being acquired. Note that poor ECG gating does not induce respiratory-induced motion artifacts. If breath-holding throughout the data collection process is adequate but ECG gating is poor, the heart alone will show motion-induced artifacts and the chest wall will appear static (Figure 12.4).


Figure 12.4 ECG gating mistriggers. A, Four-chamber view of the heart with gating mistriggers. Mixing of different cardiac views from across the cardiac cycle produces motion blurring of the heart only. B, Correctly gated 4-chamber view.

Figure 12.4 ECG gating mistriggers. A, Four-chamber view of the heart with gating mistriggers. Mixing of different cardiac views from across the cardiac cycle produces motion blurring of the heart only. B, Correctly gated 4-chamber view.

Solutions

  • Check all lead voltage waveforms for optimal ECG signal.

  • Recheck the placement of electrodes. Check for adequate electrical contact between the electrode and skin surface by using appropriate skin preparation (eg, abrasive gel).

  • Decrease the tolerance for ECG arrhythmias. This reduces the range of heart rates over which data are collected, eliminating the effect of collecting data at different phases of the cardiac cycle because of irregular heart rates or false ECG triggers. This can increase the image acquisition time because of the extra time required to complete data acquisition.

  • Increase the views per segment and, if necessary, decrease the number of cardiac phases, which will decrease the number of R-R intervals over which data are collected and potentially the amount of mistriggered data.

  • Check to ensure the ECG cable travels as closely as possible to the center of the MR scanner bore as it exits.

  • Choose peripheral pulse gating (eg, using a pulse plethysmograph).

Respiratory (Breathing) Motion Artifact

Source

Respiratory motion artifacts occur if patients are unable to hold their breath and resume breathing during the data acquisition process. Movement of the heart through the imaging plane due to diaphragm motion induces blurring, ghosting, and volume averaging of the heart (Figure 12.5). These effects can decrease image quality and the accuracy of quantitative measures such as ejection fraction and myocardial mass. This motion is distinguishable from ECG gating artifacts because of the presence of chest wall motion–induced ghosting.


Figure 12.5 Respiratory (breathing) motion. Short-axis images of the heart during a free-breathing (A) and a breath-hold (B) ECG-gated acquisition. Although ECG gating in this imaging sequence was successful, respiratory motion severely degrades image quality. Movement of the diaphragm results in movement of the cardiac anatomy through the imaging slice, resulting in blurring and ghosting of the cardiac anatomy.

Figure 12.5 Respiratory (breathing) motion. Short-axis images of the heart during a free-breathing (A) and a breath-hold (B) ECG-gated acquisition. Although ECG gating in this imaging sequence was successful, respiratory motion severely degrades image quality. Movement of the diaphragm results in movement of the cardiac anatomy through the imaging slice, resulting in blurring and ghosting of the cardiac anatomy.

Solutions

  • Decrease the imaging time, if possible, so that the patient can achieve a breath-hold throughout the imaging acquisition. Parallel imaging techniques are useful in this regard. Another approach is to decrease the number of phase-encoding steps of the imaging sequence. If a perfusion sequence is being used, decreasing the number of phases will also decrease imaging time. Increasing views per segment reduces acquisition time, but at the expense of temporal resolution.

  • In some instances, use navigator echoes or respiratory bellows, which allow free breathing during data acquisition, particularly for non-cine sequences such as T1-weighted black blood imaging.

  • Use single-shot techniques so that the patient can free breathe during data acquisition.

  • Coach the patient before the breath-hold acquisition.

Flow-Related Artifacts

Source

Blood flow into and out of the imaging volume during data acquisition results in modulation of the magnetization of the blood throughout the imaging sequence. This modulation produces replication and blurring of the signal along the phase-encoding direction of the image (Figure 12.6). This effect is most apparent for balanced steady-state free precession (SSFP) sequences and is exacerbated at increased pulse repetition time (TR) and higher field strength (3.0T). Longer TR allows more time for the blood to flow out of the volume and accumulate phase, and higher field strength increases the susceptibility variation, as measured in hertz.


Figure 12.6 Flow-related artifacts. A, Image shows fully magnetized blood in the aorta entering the imaging slice during a balanced SSFP gradient echo imaging sequence. B, Image shows the same slice but during late diastole when flow is at a minimum.

Figure 12.6 Flow-related artifacts. A, Image shows fully magnetized blood in the aorta entering the imaging slice during a balanced SSFP gradient echo imaging sequence. B, Image shows the same slice but during late diastole when flow is at a minimum.

Solutions

  • Use the shortest possible TR for balanced gradient echo sequences.

  • Swap the phase- and frequency-encoding directions to change the direction of artifact propagation.

  • Choose a spoiled rather than balanced gradient echo sequence.

  • Image at a lower field strength (1.5T vs 3.0T)

  • For non-cine acquisitions, choose a more quiescent portion of the cardiac cycle for data collection (diastole vs systole).

Susceptibility-Induced Signal Loss

Source

Differences in magnetic susceptibility of tissues or implanted materials (eg, metals such as sternal wires or stents) distort the magnetic field around the tissue interface or object, producing image distortion and signal loss (Figure 12.7).


Figure 12.7 Susceptibility-induced signal loss. Multiple short-axis views of the heart showing signal loss and distortion around sternal wires (arrows).

Figure 12.7 Susceptibility-induced signal loss. Multiple short-axis views of the heart showing signal loss and distortion around sternal wires (arrows).

Solutions

  • Use linear or higher-order shimming of the volume around the region of differing tissue susceptibilities. This can improve the main magnetic field homogeneity and sometimes decrease the artifact. Localized shimming (eg, a manually selected shim box) is particularly useful.

  • Increase the receiver bandwidth of the imaging sequences or choose spin echo–based sequences for imaging around metal implants when appropriate.

Banding on Balanced SSFP Images

Source

Balanced SSFP imaging requires that the net area on each gradient axis be zero during any TR interval. Spatially varying magnetic field inhomogeneities introduce net phase accrual and violate this condition. This results in varying bright and dark bands across the image (Figure 12.8). The source of this inhomogeneity can be the physical design limitations of the main magnetic field or susceptibility-induced field variations from different tissue interfaces (eg, lung to liver to heart) or implanted devices (eg, stents).


Figure 12.8 Banding on balanced SSFP images. A and C, Balanced SSFP gradient echo images demonstrating banding and signal loss (arrows) due to magnetic field inhomogeneities. B and D, Spoiled gradient echo images without banding.

Figure 12.8 Banding on balanced SSFP images. A and C, Balanced SSFP gradient echo images demonstrating banding and signal loss (arrows) due to magnetic field inhomogeneities. B and D, Spoiled gradient echo images without banding.

Solutions

  • Minimize the TR. The spatial period of the banding is proportional to the inhomogeneity of the main magnetic field and the inverse of the TR of the imaging sequence. Decreasing the TR will increase the separation of the bands and potentially shift these artifacts outside the anatomy of interest. This is especially important at field strengths greater than 1.5T (eg, 3.0T). For field strengths of 3.0T, the TR should not exceed 3.5 milliseconds.

  • If decreasing the TR is not possible (ie, the TR is already at its minimum value), improve the homogeneity of the main magnetic field by linear or higher-order shimming over the localized volume of the heart to decrease the artifact.

  • Switch to another type of pulse sequence such as a spoiled gradient echo; however, signal-to-noise ratio may decrease.

  • Change the resonant frequency of the receiver, thereby shifting the banding artifact so that the region of signal loss is outside the critical anatomical region of interest.

Image-Based Parallel Imaging Reconstruction Artifact

Source

Image-based parallel imaging methods such as SENSE reduce the imaging FOV in the phase-encoding direction. This decreases the total acquisition time by reducing the number of phase-encoding steps but also causes image aliasing (ie, wraparound artifact) (Figure 12.9). To unwrap the aliasing within the image, a map of the RF coil sensitivity throughout the imaging volume must be used. This is often referred to as a calibration scan and is obtained before or during acquisition of the accelerated image data (so-called autocalibration). Artifacts arise when:

  1. 1. The calibration scan does not cover the entire anatomy in the volume scanned. Regions of low signal in the calibrated region, such as in the lungs, can produce similar artifacts.

  2. 2. The reconstructed (ie, full) FOV is smaller than the object. This is particularly problematic for autocalibrated scans, which can derive their sensitivity maps from the accelerated image data, thereby producing erroneous coil-sensitivity maps.

  3. 3. The acceleration factor is too high (equal to or greater than the number of coil elements in the phased-array RF coil). In this case, the unwrapping algorithm is ill-conditioned (number of aliases greater than the number of coil elements in the phased-array RF coil) and the reconstruction algorithm will also produce image artifacts.


Figure 12.9 Image-based parallel imaging reconstruction artifact. Short-axis spoiled gradient echo images acquired with parallel imaging (SENSE) techniques (A, C, and E), and full-acquisition short-axis views (B, D, and F). Insufficient coverage of the imaging volume by the calibration scan results in reconstruction errors in the reconstructed, accelerated data (arrows).

Figure 12.9 Image-based parallel imaging reconstruction artifact. Short-axis spoiled gradient echo images acquired with parallel imaging (SENSE) techniques (A, C, and E), and full-acquisition short-axis views (B, D, and F). Insufficient coverage of the imaging volume by the calibration scan results in reconstruction errors in the reconstructed, accelerated data (arrows).

Solutions

  • Ensure that the calibration scan field more than covers the object imaged.

  • Ensure that the reconstructed (ie, full) FOV is greater than the object’s extent in the phase-encoding direction.

  • Decrease the acceleration factor to be equal to or less than the number of RF coil elements.

  • Acquire a fully sampled MR dataset but with a decreased FOV (and scan time). This intentionally generates image aliasing but with the aliased portions of the image being outside the anatomy of interest (eg, the heart).

  • Use properly designed RF coils with a higher number of elements, and/or an MR system with an increased number of channels, which allows the use of a higher acceleration factor without serious artifacts.

k-Space–Based Parallel Imaging Artifact

Source

k-Space–based methods such as GRAPPA and ARC do not require the use of a separate calibration scan. Scan times are decreased by effectively skipping k-space lines, except for the central portion of k-space. Missing k-space data are synthesized from adjacent k-space lines using weighting functions derived from the fully sampled portion of k-space. Artifacts occur when synthesized (ie, missing) lines of k-space are incorrectly reconstructed because of errors in the weighting functions (Figure 12.10). Incorrect weighting functions are the result of an insufficient number of fully sampled, central lines of k-space from which these functions are derived. Because the weighting functions apply to all missing lines of k-space data, artifacts appear as increased noise throughout the image and are not spatially localized, as is the case with sensitivity maps in image-based approaches. Aliasing artifacts can also be observed due to incorrect reconstruction of missing data.


Figure 12.10 k-Space–based parallel imaging artifact. A, Subtle aliasing of the FOV in the phase-encoding direction (left-to-right) can be seen in the center of the image (arrows) due to incorrect synthesis of skipped k-space data. B, Short-axis, full-acquisition, balanced SSFP image. C, Highly accelerated k-space–based parallel imaging acquisition demonstrating noise amplification throughout the image introduced by synthesis of missing k-space lines.

Figure 12.10 k-Space–based parallel imaging artifact. A, Subtle aliasing of the FOV in the phase-encoding direction (left-to-right) can be seen in the center of the image (arrows) due to incorrect synthesis of skipped k-space data. B, Short-axis, full-acquisition, balanced SSFP image. C, Highly accelerated k-space–based parallel imaging acquisition demonstrating noise amplification throughout the image introduced by synthesis of missing k-space lines.

Solutions

  • Decrease the acceleration factor, thereby increasing the number of sampled lines of k-space that are used to generate the weighting functions used in the reconstruction of missing k-space lines.

  • Acquire a fully sampled MR dataset but with a reduced FOV (and scan time), thereby intentionally generating image aliasing but with the aliased portions of the image being outside the anatomy of interest (eg, the heart).

  • Use properly designed RF coils with a higher number of elements, and/or an MR system with an increased number of channels, which allows the use of a higher acceleration factor without serious artifacts.

Perfusion Dark-Rim Enhancement Artifact

Source

Perfusion dark-rim enhancement is a loss of signal within the myocardium during first-pass perfusion imaging that can mimic myocardial perfusion defects (Figure 12.11). Several possible sources have been identified, including Gibbs ringing, motion, k-space sampling strategies, and a too-high concentration of contrast agent resulting in susceptibility-induced signal loss. The artifact is transient and typically disappears as the peak of the bolus passes through the ventricles, as opposed to a true perfusion defect, which will be present throughout the first pass of the contrast agent. The artifact can be verified by the lack of signal enhancement at the site of the signal void on late gadolinium-enhancement (LGE) images, as shown in Figure 12.11.


Figure 12.11 Perfusion dark-rim enhancement artifact. Example of the dark-rim enhancement artifact (arrows) on short-axis views of the heart during first-pass perfusion imaging. The first 3 images show baseline (A), arrival of the bolus (B), and washout (C) of the contrast agent. During the peak bolus phase, the dark-rim enhancement artifact is manifested as a dark line along the endocardial border of the septal wall of the left ventricle and can be mistaken for a perfusion defect. D, The artifact is present for only a few cardiac phases and therefore is not present in the last perfusion image. The LGE image of the same slice does not show evidence of an infarction.

Figure 12.11 Perfusion dark-rim enhancement artifact. Example of the dark-rim enhancement artifact (arrows) on short-axis views of the heart during first-pass perfusion imaging. The first 3 images show baseline (A), arrival of the bolus (B), and washout (C) of the contrast agent. During the peak bolus phase, the dark-rim enhancement artifact is manifested as a dark line along the endocardial border of the septal wall of the left ventricle and can be mistaken for a perfusion defect. D, The artifact is present for only a few cardiac phases and therefore is not present in the last perfusion image. The LGE image of the same slice does not show evidence of an infarction.

Solutions

  • Decrease the injection rate of contrast agent. This broadens the contrast bolus and decreases first-pass image contrast.

  • Increase the number of phase-encoding steps. This will increase the spatial resolution, decreasing the potential for Gibbs ringing and susceptibility-induced signal loss. This will also increase the acquisition time, which can then be decreased by using k-space–based parallel imaging approaches.

  • Zero-fill the k-space data before Fourier transformation. This results in smaller pixel dimensions by means of interpolation and can more clearly depict Gibbs ringing artifacts, reducing their dependence on the specific position of the endocardial wall.

Lightning Flash Artifact

Source

Most MR imaging techniques require establishment of the steady state (ie, the dynamic balance between longitudinal signal recovery and the creation of transverse magnetization after an RF excitation pulse). Imaging under non–steady-state conditions results in modulation of the magnitude of the frequency (k-space) data and subsequent image artifacts. For gradient echo cine sequences, tissues with short T1 relaxation time will recover faster and contribute more signal under non–steady-state conditions. This typically occurs due to irregular heart rates. For cine acquisitions, non–steady-state images will appear brighter (higher overall signal) than images acquired under steady-state conditions. When reviewed as a cine loop, non–steady-state images will appear as a “flash” of increased signal in 1 portion of the cine loop (Figure 12.12).


Figure 12.12 Lightning flash artifact. Example of loss of the steady-state condition during acquisition of a short-axis cine image series. Because of the short T1 of the lipid signal, ghosting artifacts are apparent and, when viewed as a cine loop, appear as a bright or lightning flash. A and B, Short-axis views acquired during early systole before the steady state has been established. Ghostlike signal in the background and in the lungs is an artifact due to the segmented nature of the data acquisition scheme and the transition from non–steady state to steady state of the signal. C, The steady state is established, and related artifacts are absent, in an image acquired at a later phase of systole.

Figure 12.12 Lightning flash artifact. Example of loss of the steady-state condition during acquisition of a short-axis cine image series. Because of the short T1 of the lipid signal, ghosting artifacts are apparent and, when viewed as a cine loop, appear as a bright or lightning flash. A and B, Short-axis views acquired during early systole before the steady state has been established. Ghostlike signal in the background and in the lungs is an artifact due to the segmented nature of the data acquisition scheme and the transition from non–steady state to steady state of the signal. C, The steady state is established, and related artifacts are absent, in an image acquired at a later phase of systole.

Solutions

  • Use a pulse sequence that continuously applies RF pulses during the entire R-R interval, including the window when ECG triggers are detected. This maintains the steady state throughout the data collection process. This is particularly important if the patient has an irregular heartbeat. If the patient’s heart rate is regular but the imaging sequence is missing ECG triggers, then several of the methods for improving gating reliability should be implemented. The arrhythmia rejection window (if applicable) should be increased to reject data outside the average heart rate of the patient.

Susceptibility-Induced Fat Saturation Failure

Source

Chemical fat saturation techniques apply an RF pulse that is tuned to the resonant frequency of lipids within the imaging volume. Magnetic field inhomogeneities induced by susceptibility differences alter the resonant frequency of not only water but also lipids. If the susceptibility-induced frequency shift is large enough, the resonant frequency of the lipid signal can shift outside the pass band of the fat saturation RF pulse. In certain circumstances the resonant frequency of water can shift into the pass band of the RF pulse. Under these conditions, the water signal is suppressed and the fat signal remains unaffected. Figure 12.13 demonstrates this effect in the thorax, with the signal from water being suppressed, creating a signal void about the ascending aorta, while the lipid signal remains unaffected.


Figure 12.13 Susceptibility-induced fat saturation failure. Images demonstrate susceptibility-induced alteration of the main magnetic field within the thorax at the level of the aortic arch. A, In this fat-saturated, T1-weighted volumetric acquisition, susceptibility differences between the lungs, anterior chest wall, and great vessels are large enough to shift the resonant frequency of tissue within the ascending aorta into the saturation band of the fat saturation pulse, which results in signal loss within this vessel (arrow). The use of iterative Dixon-based methods results in water-only (B) and fat-only (C) images, which are insensitive to off-resonance effects.

Figure 12.13 Susceptibility-induced fat saturation failure. Images demonstrate susceptibility-induced alteration of the main magnetic field within the thorax at the level of the aortic arch. A, In this fat-saturated, T1-weighted volumetric acquisition, susceptibility differences between the lungs, anterior chest wall, and great vessels are large enough to shift the resonant frequency of tissue within the ascending aorta into the saturation band of the fat saturation pulse, which results in signal loss within this vessel (arrow). The use of iterative Dixon-based methods results in water-only (B) and fat-only (C) images, which are insensitive to off-resonance effects.

Solutions

  • Image the anatomy of interest using iterative Dixon imaging techniques. These techniques acquire image data at 2 or 3 different echo times, followed by a special iterative reconstruction that attempts to correct for magnetic field inhomogeneities to improve the ability to separate fat and water signals into fat-only and water-only images. The separate fat and water images are absent of any signal voids and accurately reproduce the anatomy of interest.

  • In the absence of Dixon-based fat-water imaging methods, the homogeneity of the main magnetic field may be improved by performing so-called small-volume linear or higher-order shimming. Localized, small-volume shimming allows a volume of interest to be identified about the anatomy of interest that excludes the remainder of the imaging volume. The spatial encoding gradients are used to attempt to minimize the inhomogeneity within the volume and hence reduce susceptibility-induced off-resonance effects. Higher-order shimming involves the use of special shimming coils and circuits that create magnetic fields described by polynomial functions of spatial position. This approach has not yet been commonly used.

Incomplete Myocardial Suppression on LGE Imaging

Source

Incorrect choice of inversion time on LGE imaging results in incomplete suppression of the signal from normal myocardium. Incomplete suppression of normal myocardium decreases overall image contrast and decreases conspicuity of infarcted regions (Figure 12.14).


Figure 12.14 Incomplete myocardial suppression on LGE imaging. This artifact is the effect of incorrect choice of inversion time. A, Single-shot, long-axis, magnitude-only, LGE image. Incorrect inversion time results in incomplete suppression of normal myocardium and poor normal-to-infarcted myocardial contrast. B, Phase-sensitive LGE image with the same inversion time demonstrates improved normal-to-infarcted myocardial contrast.

Figure 12.14 Incomplete myocardial suppression on LGE imaging. This artifact is the effect of incorrect choice of inversion time. A, Single-shot, long-axis, magnitude-only, LGE image. Incorrect inversion time results in incomplete suppression of normal myocardium and poor normal-to-infarcted myocardial contrast. B, Phase-sensitive LGE image with the same inversion time demonstrates improved normal-to-infarcted myocardial contrast.

Solutions

  • Check the inversion time by running T1 mapping or T1 scout sequences. As a result of the wash-in and wash-out of contrast agent into normal and infarcted myocardium, a delay between running of the T1 mapping and LGE sequences can result in the choice of an inversion time that is no longer accurate at the time of imaging. Rechecking the inversion time by repeating the mapping routine can correct this discrepancy, as long as sufficient contrast is still within the blood pool and myocardium.

  • Use phase-sensitive LGE imaging methods. This method preserves the sign of the MR signal, is less sensitive to incorrect choice of inversion time, and provides improved normal myocardium-to-infarction contrast. However, phase-sensitive images have a lower signal-to-noise ratio than magnitude-only methods.

Suggested Reading

Anatomy and Physiology of the Heart

Marieb EN. Essentials of human anatomy and physiology. 7th ed. San Francisco (CA): Benjamin Cummings; c2003.Find this resource:

    Cardiac MR

    Bogaert J, Dymarkowski S, Taylor AM. Clinical cardiac MRI: with interactive CD-ROM. Berlin: Springer; c2005.Find this resource:

    Lee VS. Cardiovascular MRI: physical principles to practical protocols. Philadelphia (PA): Lippincott Williams & Wilkins; c2006.Find this resource:

      ECG Physiology

      Hobbie RK, Roth BJ. Intermediate physics for medicine and biology. 4th ed. New York (NY): Springer; c2007.Find this resource:

        Malmivuo J, Plonsey R. Bioelectromagnetism: principles and applications of bioelectric and biomagnetic fields. New York (NY): Oxford University Press; c1995.Find this resource:

        Gradient Decay or Fall-Off

        Rangwala N, Zhou XJ. Reduction of fast spin echo cusp artifact using a slice-tilting gradient. Magn Reson Med. 2010 Jul;64(1):220–8.Find this resource:

        Miscellaneous

        Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, et al; American Heart Association Writing Group on Myocardial Segmentation and Registration for Cardiac Imaging. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Int J Cardiovasc Imaging. 2002 Feb;18(1):539–42.Find this resource:

        MR Acronyms

        Boyle GE, Ahern M, Cooke J, Sheehy NP, Meaney JF. An interactive taxonomy of MR imaging sequences. Radiographics. 2006 Nov-Dec;26(6):e24.Find this resource:

        Brown MA, Semelka RC. MR imaging abbreviations, definitions, and descriptions: a review. Radiology. 1999 Dec;213(3):647–62.Find this resource:

        Nitz WR. MR imaging: acronyms and clinical applications. Eur Radiol. 1999;9(5):979–97.Find this resource:

        Sprung K. Basic techniques of cardiac MR. Eur Radiol. 2005 Feb;15 Suppl 2:B10–6.Find this resource:

        MR Physics and Pulse Sequences

        Bernstein MA, King KF, Zhou XJ. Handbook of MRI pulse sequences. Amsterdam: Elsevier Academic Press; c2004.Find this resource:

          Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic resonance imaging: physical principles and sequence design. New York (NY): J Wiley & Sons; c1999.Find this resource:

            Perfusion Dark-Rim Enhancement Artifact

            Ferreira P, Gatehouse P, Kellman P, Bucciarelli-Ducci C, Firmin D. Variability of myocardial perfusion dark rim Gibbs artifacts due to sub-pixel shifts. J Cardiovasc Magn Reson. 2009 May 27;11:17.Find this resource:

            Vector ECG Gating

            Chia JM, Fischer SE, Wickline SA, Lorenz CH. Performance of QRS detection for cardiac magnetic resonance imaging with a novel vectorcardiographic triggering method. J Magn Reson Imaging. 2000 Nov;12(5):678–88.Find this resource:

            Fischer SE, Wickline SA, Lorenz CH. Novel real-time R-wave detection algorithm based on the vectorcardiogram for accurate gated magnetic resonance acquisitions. Magn Reson Med. 1999 Aug;42(2):361–70.Find this resource: