|Home | About | Journals | Submit | Contact Us | Français|
Delayed enhancement cardiac magnetic resonance (DE-CMR) imaging is used increasingly to identify and quantify focal myocardial scar. Our objective is to describe factors used in the interpretation of DE-CMR images and to highlight potential pitfalls and artefacts that mimic myocardial scar. Inversion recovery gradient recalled echo sequence is commonly accepted as the standard of reference for DE-CMR. There are also alternative sequences that can be performed in a single breath-hold or with free breathing. Radiologists need to be aware of factors affecting image quality, and potential pitfalls and artefacts that may generate focal hyperintense areas that mimic myocardial scar.
Delayed enhancement cardiac magnetic resonance (DE-CMR) imaging can identify the presence, location and extent of myocardial scar (dense myocardial fibrosis) due to ischaemic and non-ischaemic heart diseases. This technique is also called late gadolinium-enhanced MRI (LGE-MRI), myocardial delayed enhanced (MDE) MRI and delayed hyperenhanced MRI (DHE-MRI) in the literature. DE-CMR has been shown to be more sensitive than other imaging methods in detecting small subendocardial infarctions  and has recently been used in population-based studies to assess the presence of myocardial scar [2,3]. The technique is relatively straightforward to implement, and images can be interpreted visually for the presence of myocardial scar (hyperenhanced region) without the need for post-processing.
This article is intended to describe quality control procedures, factors affecting image quality, assessment of myocardial scar, and potential pitfalls and artefacts causing focal hyperintensities that mimic myocardial scar in delayed enhancement imaging.
DE-CMR is typically obtained using a segmented inversion recovery gradient recalled echo sequence (IR-GRE) with breath-holds, which is accepted as the standard of reference. Each slice is acquired in one breath-hold 10–20 minutes after intravenous administration of 0.1–0.2 mmol kg−1 gadolinium-based contrast agent. This sequence applies an 180° inversion pulse to null the signal of normal myocardium and enhance the signal difference between normal and scarred myocardium. The time between inversion recovery pulse and image acquisition is the “inversion time” or TI (Figure 1). The inversion time needs to be optimized for each case individually before image acquisition. With optimal settings, the normal myocardium appears nulled or “black” and myocardial scar areas appear enhanced or “bright”. The mechanism underlying delayed enhancement in the chronic disease setting is increased gadolinium concentration in expanded interstitial spaces because of replacement of non-viable myocytes with highly collagenous scar tissue (dense myocardial fibrosis) .
Delayed enhancement images are usually acquired in a stack of 10–12 short-axis slices in combination with a horizontal and a vertical long-axis image perpendicular to the short axis. Slice thickness is 6–8 mm. The gap between slices is 2–4 mm. The in-plane spatial resolution is typically 1.2–1.8 mm. The field of view is 300–380 mm and should be minimized to improve the spatial resolution while avoiding wrap-around artefact . Images are typically obtained with an electrocardiogram (ECG) delay such that mid-diastolic imaging is performed. Obtaining delayed enhancement images at the same spatial location as short-axis cine images is desirable because it enables comparison between suspected scar and local wall motion. Advances in pulse sequences allow us to acquire the entire heart in a single breath-hold with inversion recovery prepared steady-state free precession (IR-SSFP) sequence or with rapid three-dimensional (3D) inversion recovery gradient recalled echo sequences. Navigator gated 3D acquisitions cover the entire heart during free breathing. All these sequences can be acquired with phase-sensitive inversion recovery reconstruction. Major characteristics of alternative pulse sequences are compared with the standard reference sequence in Table 1.
The following factors should be checked before interpreting images to evaluate the diagnostic quality of images.
The fibrous skeleton of the heart (e.g. mitral valve annulus) is often hyperintense because of gadolinium accumulation in these highly collagenous areas. Heart valves also faintly enhance (Figure 4).
If the inversion time is slightly shorter than optimal inversion time, normal myocardium appears grey in between hypointense endo- and epicardial lines (Figure 5a).
If very long inversion time is used, the image contrast is reduced. Images acquired with suboptimal inversion time may lead to inaccurate assessment of scar size . An IR-SSFP look-locker sequence is used to select the optimal inversion time visually for maximising the signal intensity between normal and scarred myocardium  (Figure 6).
Since each slice can be obtained in one breath-hold with the IR-GRE sequence, the TI time should be slightly lengthened as the delay after contrast increases. Another way to achieve a consistent contrast over a wide range of inversion times is to apply phase-sensitive reconstruction. This method uses phase information to correct magnitude images when TI times are incorrectly determined  (Figure 5a,b). The phase-sensitive reconstruction method also reduces the variation in apparent infarct size. In patients with breath-holding difficulty, DE-CMR can be performed using a nominal inversion time value in combination with IR-SSFP sequence which will obviate the need for extra breath-hold for the look-locker sequence to optimise inversion time. A disadvantage of phase-sensitive inversion recovery images is that noise may be more prominent in the image.
We interpret a hyperintense region as definite when it is confirmed either in two adjacent short-axis images or in one short-axis image and a long-axis image at a corresponding location. Confirmation of scar in two different images avoids misinterpretation of artefacts or image features that may be due to motion artefacts, morphological variations or septal clefts. However, very small focal myocardial scar areas might be missed since slice thickness is 6–8 mm.
Myocardial scar size in DE-CMR can be quantified by automated or manual segmentation methods. These methods are reader dependent for delineation of endocardium and epicardium to calculate left ventricular (LV) mass and the scar size as a percentage of LV mass. For calculation of myocardial scar size, hyperenhanced areas are delineated by a reader using the manual segmentation method, which is observer dependent and time-consuming. Automated segmentation methods use either a threshold of signal intensity from 2 to 6 standard deviations above the remote myocardium or a threshold of full-width at half-maximum to calculate infarct size.
Alternatively, visual scoring is a fast and semi-quantitative method to detect infarct size based on 17 segments of the American Heart Association heart model, which can be used in daily clinical practice. In this method, a visual score ranging from 0 to 4 was attributed to each of the 17 segments according to the transmural extent of the hyperenhancement (0, no hyperenhancement; 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, 75–100% transmurality). All scores are summed and divided by the maximal possible score (17×4=68) and then multiplied by 100 to express global infarct size as a percentage .
The hyperintense blood between myocardium and papillary muscles can mimic myocardial scar (Figure 7).
Distinguishing small subendocardial myocardial infarctions from blood pool can also be difficult since both are hyperintense in the image. As an aid, myocardial thickness and morphology can be compared with cine MRI to help determine whether the enhancement occurs in the myocardium or in the LV cavity (Figure 8).
Epi- and pericardial fat often show high signal intensity using the inversion recovery technique. This can mimic epicardial hyperenhancement (Figure 9). Checking the continuity of myocardial scar area in adjacent slices and in other imaging planes is helpful to identify epicardial fat.
Partial volume effects typically cause problems near the aortic outflow tract (Figure 10a–c) and epicardial locations due to right ventricular septomarginal trabeculation, as well as the right ventricular insertion site near the apex (Figure 11a–c). Myocardial hyperenhancement is considered to represent myocardial scar if it can be seen continuously in two consecutive slices (Figure 11d–g).
Myocardial clefts present as a discrete pouch of blood signal penetrating more than half the thickness of the myocardial wall in the basal inferior wall of the left ventricle or the mid- to apical segments of the interventricular septum. They typically narrow or occlude during systole . The blood inside the cleft can simulate myocardial scar in delayed enhancement images. Differentiation can be accomplished by evaluating the pouch and wall morphology on cine MRI (Figure 12).
Myocardial clefts also need to be differentiated from non-compacted myocardium that is characterized by the diastolic ratio of non-compacted to compact layers of myocardium >2.3 at diastole .
The left anterior descending artery gives rise to a variable number of septal perforator branches. Approximately 15% of subjects have a large first septal perforator branch, which may mimic linear hyperenhancement in a short-axis delayed enhancement image at the base of the heart (Figure 13).
Stomach motion artefacts simulate myocardial scar in one short axis slice, especially on the right ventricular insertion point. They reproduce the size and shape of the stomach along the phase encoding direction. This finding is the important clue to differentiate artefact from myocardial scar (Figure 15).
In cardiac amyloidosis, the accumulation of amyloid proteins in interstitial space increases contrast accumulation diffusely. Thus, detection of the optimal inversion time to suppress the normal myocardial signal is typically difficult. A pronounced subendocardial, global or diffuse increase in signal intensity, which does not match any coronary artery territory, is the characteristic enhancement pattern for cardiac amyloidosis  (Figure 16).
DE-CMR represents an emerging standard for identification and quantification of myocardial scar/fibrosis. The IR-GRE sequence is commonly accepted as the standard of reference. There are alternative sequences that can be performed in a single breath-hold or with free breathing in patients with breath-holding difficulty. Radiologists need to be aware of potential pitfalls and artefacts that can be seen as hyperintense areas in delayed enhancement images and may otherwise mimic myocardial scar. It is important to confirm the presence of a hyperintense area in more than one slice (two adjacent short-axis slices or one short-axis and one long-axis image at a corresponding location) and to assess the corresponding cine images when interpreting DE-CMR of myocardial scar. These cross-checks reduce the opportunity for misinterpretation; however, despite these improvements, current DE-CMR resolution is still a limitation in the detection of very small myocardial scars.