Visualization of cerebral microbleeds with dual-echo T2*-weighted magnetic resonance imaging at 7.0 T

Authors

  • Mandy M.A. Conijn MD,

    Corresponding author
    1. Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands
    2. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, Netherlands
    • University Medical Center Utrecht, Department of Radiology (Hpn E 01.132), PO Box 85500, 3508 GA Utrecht, the Netherlands
    Search for more papers by this author
  • Mirjam I. Geerlings PhD,

    1. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, Netherlands
    Search for more papers by this author
  • Peter R. Luijten PhD,

    1. Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands
    2. Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands
    Search for more papers by this author
  • Jaco J.M. Zwanenburg PhD,

    1. Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands
    2. Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands
    Search for more papers by this author
  • Fredy Visser,

    1. Philips Healthcare, Best, Netherlands
    Search for more papers by this author
  • Geert Jan Biessels MD, PhD,

    1. Department of Neurology, University Medical Center Utrecht, Utrecht, Netherlands
    Search for more papers by this author
  • Jeroen Hendrikse MD, PhD

    1. Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands
    Search for more papers by this author

Abstract

Purpose:

To assess the visualization of cerebral microbleeds with dual echo T2*-weighted imaging at 7.0 T magnetic resonance imaging (MRI).

Materials and Methods:

Ten consecutive participants (eight men, two women, mean age 54 ± 12 years) with vascular disease or risk factors from the second manifestations of arterial disease (SMART) study were included. Dual-echo T2*-weighted scans (echo time: 2.5/15.0 msec) were made for all participants at 7.0 T MRI. The number of visible microbleeds and the diameter of the microbleeds were recorded on minimal intensity projection images of both echoes.

Results:

The first echo image shows dark microbleeds against a homogeneous, more hyperintense signal of the brain tissue without contrast for veins and basal ganglia. In eight patients microbleeds were observed, with a total of 104 microbleeds. Of these, 88 (84.6%) were visible on the first and 102 (98.0%) on the second echo. The mean diameter of the microbleeds was 1.24 mm for the first echo and 2.34 mm for the second echo.

Conclusion:

T2*-weighted imaging at two echo times at 7.0 T combines the advantages of the first and second echo. Microbleeds visible on the first echo show large contrast with the surrounding tissue, even in the presence of paramagnetic ferritin. The second echo enables visualization of smaller microbleeds than the first echo. J. Magn. Reson. Imaging 2010;32:52–59. © 2010 Wiley-Liss, Inc.

THE INTEREST IN CEREBRAL MICROBLEEDS is growing, as is shown by the increasing number of studies investigating the prevalence and the clinical relevance of these lesions. The studies done until now have increased the understanding of cerebral microbleeds. They have shown that cerebral microbleeds appear to be direct markers of vascular disease, as they are associated with hypertensive vasculopathy and cerebral amyloid angiopathy, and also with white matter lesions and lacunar infarcts (1–6). The finding of microbleeds can be of major importance in patients with ischemic stroke receiving anticoagulation, as the presence of microbleeds might indicate a higher risk of future intracerebral hemorrhage, although this is still under debate (7–12). Furthermore, some studies have shown that microbleeds are associated with cognitive impairment, functional dependence, or death (3, 13, 14). However, as some of these associations are still under debate and the clinical implication of microbleeds is not completely clear, more and larger studies are needed to obtain more information about microbleeds (3).

The number of microbleeds detected, and with that the study sensitivity, differs strongly between different image protocols and different field strengths. This can be illustrated by the varying prevalence of microbleeds found in different studies. The prevalence found in population-based studies ranges from 4.7%–23.5% (15, 16). An optimal imaging protocol for the visualization of microbleeds is hard to define, because there is no reference test available for the detection of these lesions.

Microbleeds consist of hemosiderin deposits that are paramagnetic, due to the presence of paramagnetic iron. This induces a susceptibility effect on the magnetic resonance imaging (MRI) scan, which leads to a fast decay of the local T2*-weighted MRI signal because of a local inhomogeneity of the field induced by the internal magnetization of microbleeds. T2*-weighted gradient-recalled echo (GRE) has been shown to be highly sensitive to this susceptibility effect, and is therefore very sensitive to microbleeds (2, 3, 17). As the effect scales with the magnetic field, the detection of microbleeds is substantially improved at ultrahigh field strengths. Moreover, increased signal-to-noise ratio (SNR) can be used for higher spatial resolution and increased conspicuity of small hemosiderin deposits that may be obscured by partial volume effects at lower field strengths that operate at lower spatial resolution (18).

Besides the field strength, the echo time is also an important parameter for the visualization of microbleeds. A longer echo time gives more time for dephasing, which enhances the susceptibility effect. This so-called blooming effect causes the microbleeds to appear as hypointense spots that are larger than the actual size. It has been shown that prolonging the echo time leads to an increase in diameter of these lesions and also detection of an increased number of microbleeds (19). However, microbleeds can also be obscured by overlapping structures with a high susceptibility effect, like veins (deoxyhemoglobin) or the basal ganglia (ferritin deposition). With a longer echo time, these structures will also increase in size and become more dominant, making it harder to distinguish the microbleeds.

The magnetic characteristics of paramagnetic substances such as deoxyhemoglobin, ferritin, and hemosiderin depend largely on the amount of iron present. Deoxyhemoglobin in veins is less paramagnetic than ferritin depositions in the basal ganglia and hemosiderin depositions in microbleeds, which have strong paramagnetic iron-containing cores. Hemosiderin is a degradation product of ferritin, in which the iron cores are more closely packed (20), which explains why hemosiderin was shown to have a stronger T2 shortening effect than ferritin (21–23). As the magnetic characteristics of deoxyhemoglobin, ferritin, and hemosiderin are different, the blooming effect as a function of echo time will also be different.

With use of a shorter echo time, it may be possible to better distinguish the hemosiderin-containing microbleeds from other structures with a high susceptibility like veins and ferritin-containing basal ganglia. Therefore, a trade-off needs to be found between a long echo time that visualizes more and larger microbleeds and a short echo time that suppresses overlap from other structures with a high susceptibility and gives more SNR.

Recently, a method was introduced that uses T2*-weighted imaging using two echo times. This method was designed for simultaneous angiography and venography (24). However, the use of two echo times may also be useful for the visualization of microbleeds. Therefore, the purpose of this study was to evaluate the visualization of cerebral microbleeds with T2*-weighted imaging using two echo times at 7.0 T MR.

MATERIALS AND METHODS

Participants

For this study we included 10 consecutive participants from the second manifestations of arterial disease (SMART) study. In the SMART study all eligible patients, aged 18 to 79 years, newly referred to our hospital with symptomatic atherosclerotic disease or risk factors for atherosclerosis were included. The objectives of the SMART study are to determine the prevalence of vascular risk factors and concomitant arterial disease and to study the incidence of future cardiovascular events and its predictors in these high-risk patients (25). The SMART study and the 7.0 T imaging were approved by the Medical Ethics Committee. Written informed consent was given by all participants.

The participants were eight men and two women, with a mean age of 54 years (standard deviation [SD] 12 years). Six of the 10 participants were included in the SMART study with vascular risk factors (five diabetes mellitus, one family history of vascular disease), two participants with a stroke, one participant with angina pectoris, and one participant with peripheral artery disease.

MRI

MRI was performed with a 7.0 T whole-body system (Philips Healthcare, Cleveland, OH) using a volume transmit and 16-channel receive head coil (Nova Medical, Wilmington, MA). In all participants a dual echo T2*-weighted sequence was used with an acquired resolution of 0.35 × 0.4 × 0.6 mm3. The dual echo sequence that we investigated for the detection of microbleeds was originally designed for simultaneous arterial and venous angiography and therefore the first echo was optimized as much as possible for angiography and the second echo for venography. The echo time for the first echo was 2.5 msec for angiography and the echo time for the second echo was 15.0 msec for venography. These echo times were chosen to obtain a good balance between good background suppression for the angiogram and enough sensitivity (T2*-weighting) and SNR for the venogram. The other imaging parameters were: repetition time 20 msec and acquired matrix of 508 × 399 with 167 slices. Excitation pulses consisted of nonsaturated excitation pulses with nominal flip angle variation of 16–24° in the feet–head direction over the slab. Flow compensation was applied in three directions. For the first echo a partial echo readout was used; for the second echo a full echo readout. A fly-back gradient was applied between the two readouts. Sensitivity encoding was applied in the RL direction with an acceleration factor of 2.5. The images were reconstructed to 0.35 × 0.35 × 0.3 mm3 voxels and the built-in phase correction, partial-echo filter, and homogeneity correction of the MR system were applied during reconstruction. The imaging duration was 8 minutes 50 seconds.

Because of ongoing development at 7.0 T MR, some scans were acquired with a slightly different resolution.

Postprocessing

The postprocessing of the data was performed on the standard console. Minimum intensity projections (minIPs) were reconstructed for transversal slabs (thickness 3 mm, 2 mm overlap, 150 slices) of the first echo and the second echo. To visualize the arteries, a maximum intensity projection (MIP) was reconstructed for transversal slabs (thickness 20 mm, 18 mm overlap, 50 slices) of the first echo. A summation was made of the resulting MIP image with a minIP (thickness 10 mm, 8 mm overlap, 50 slices) of the first echo image to visualize the relation between arteries and microbleeds.

Rating of Microbleeds

Microbleeds were defined as round or ovoid black lesions on the first and/or the second echo of the T2*-weighted scan. Symmetrical areas of calcification in the basal ganglia, choroid plexus, and pineal gland were excluded, as were signal voids caused by sulcal vessels and low-signal lesions thought to be signal averaging from adjacent bone (3, 15).

The minIP images of the first and second echo were linked and analyzed simultaneously. The presence of microbleeds was analyzed on the first and second echo image by one observer. Intensity profiles of the microbleeds on the first and/or second echo were obtained, as illustrated in Fig. 1. The software program MatLab (v. 7.6, MathWorks, Natick, MA) was used to calculate the full-width-at-half-minimum (FWHM) of the profiles. We took the FWHM as an estimate of the diameter of the microbleeds as they appeared in the images.

Figure 1.

a: A horizontal line is drawn in the midline of the microbleed on a minimal intensity projection of the first echo image to obtain the intensity profile. The corresponding profile is shown in b. This profile is used to calculate the full-width-at-half-minimum as an estimate for the diameter of the microbleed.

Statistical Analysis

First, the number of microbleeds was calculated for each participant. Second, the difference in diameter between the two echoes was tested with use of the nonparametric Wilcoxon signed ranks test for all microbleeds that were visible on both echoes, as the diameters of the microbleeds were not normally distributed. The correlation between the diameter on the first and second echo was estimated with Spearman's correlation coefficient. The analysis was performed by using the statistical software package SPSS (v. 15.0 for Windows; Chicago, IL).

RESULTS

Figure 2 shows an example of the angiography on the MIP of the first echo (Fig. 2a) and the venography on the minIP of the second echo (Fig. 2d) with dual echo T2*-weighted imaging at 7.0 T. A summation of the MIP of the first echo (Fig. 2a) with the minIP of the first echo (Fig. 2b) enables visualizing the relation between the arteries and the microbleeds (Fig. 2c). The minIP of the second echo shows the relation between veins and microbleeds (Fig. 2d,e).

Figure 2.

The first and second echo image can be used to visualize different structures. a: Maximum intensity projection (MIP) of the first echo first echo image (thickness 20 mm), showing the arteries. The box indicates the area shown in panels C,E. b: minIP of the first echo (thickness 10 mm), showing a homogeneous background, without contrast for veins and basal ganglia. A small microbleed is visible as a round, black lesion (arrow). c: A summation of the MIP (A) and the minIP (B) of the first echo enables visualizing the relation between microbleeds and the arteries. d: MinIP of the second echo (thickness 10 mm), showing the veins and a small microbleed (arrow). e: Magnification of the box in D, showing the relation between the microbleed (arrow) and the veins. Note the lack of signal in the globus pallidus, which makes it difficult to observe potential microbleeds in this area on the second echo.

The distinction between veins and microbleeds can be well made with use of the minIP of the second echo. The round hypodensities that mimic microbleeds on the source data turn out to be linear structures on the minIP of the second echo that can be identified as veins. As an example in Fig. 3a, the small area of darkening indicated by the arrow is not a microbleed, which can be clearly seen from the minIP images (Fig. 3b), which show that this area corresponds with a vein.

Figure 3.

a: Several round hypodensities are seen on this source image of the second echo which are suspicious for microbleeds (arrows). b: With use of a minimal intensity projection of the second echo image (thickness 3 mm), the hypodensities that were suspicious for microbleeds turn out to be venous structures (arrows).

Of the 10 participants, microbleeds were found on the first or second echo in eight participants. In these eight participants, a total of 104 microbleeds were detected on the first or second echo, with a median of 4 microbleeds (range 1–77). Of these 104 microbleeds, 88 (84.6%) were visible on the first echo. On the second echo 102 (98.0%) microbleeds were visible. Of all the microbleeds that were visible on the second echo, 16 microbleeds were not visible on the first echo (Fig. 4). In general, these were microbleeds with a small diameter (mean diameter 0.78 mm [SD 0.18 mm] on the second echo). Two microbleeds were only visible on the first echo image, whereas on the second echo image the microbleeds could not be distinguished from surrounding tissue or veins (Fig. 5). These microbleeds had a diameter of 0.89 mm and 1.48 mm on the first echo.

Figure 4.

a: A minIP of the first echo image. The microbleeds that are visible on the minIP of the second echo image (b) are not visible on the first echo image (arrows).

Figure 5.

a: A microbleed that is clearly visible on the minIP of the first echo image (arrow). The same microbleed is hardly distinguishable from the ferritin-containing putamen on the minIP of the second echo image (b, arrow). In c there is a microbleed visible on the minIP of the first echo image, which disappears behind a venous structure on the minIP of the second echo image in d.

The 86 microbleeds that were visible on both the first and the second echo image had a mean diameter of 1.24 mm (SD 0.56) on the first echo image and 2.34 mm (SD 1.0) on the second echo. Comparison of the diameters visible on the first and second echo image in all participants showed that the microbleeds on the second echo were significantly larger than on the first echo image (P < 0.001). The growth in diameter between the two echo times is illustrated in Fig. 6. The diameters show a strong linear correlation between the first and second echo time (correlation coefficient 0.85, P < 0.001).

Figure 6.

a: The microbleed that is visible on minIP of the first echo image (arrow) appears larger on the minIP of the second echo image (b, arrow).

DISCUSSION

With this study we showed that a short echo time is beneficial to distinguish microbleeds from other structures with a high susceptibility and that with the second echo a larger number of microbleeds can be detected. The advantages of both echo times are combined in the dual-echo T2*-weighted imaging at 7.0 T.

We showed that microbleeds appear larger on the second echo than on the first echo; however, the real size of the microbleeds is not reflected in these images, due to the blooming effect. This is the case in all T2*-weighted or susceptibility-weighted images; the real size of the microbleeds can only be obtained with pathology. The increase in diameter explains why many microbleeds can be seen on the second echo image, whereas they are too small to detect on the first echo image. The long echo time gives more time for dephasing, which enlarges the susceptibility effect (3). This leads to a larger blooming effect and thus a larger diameter of the microbleeds, which is very useful for the detection of the microbleeds. From the literature it is known that the observed volume of a microbleed is proportional to the echo time (26). As the volume is proportional to the third power of the diameter, the increase in volume in our study is (2.34/1.24)3 = 6.7, which corresponds reasonably to the increase in echo time of 15/2.5 = 6. Another advantage of the second echo image is that the veins are clearly visible. This can provide information about the relation between veins and microbleeds.

However, microbleeds are often found near veins or in the basal ganglia, and, with increasing age, ferritin-bound iron accumulates in the basal ganglia. Because the susceptibility effect of hemosiderin is stronger than that of deoxyhemoglobin and ferritin (21–23), hemosiderin can already be visualized with a short echo time, whereas deoxyhemoglobin and ferritin are hardly visible on the first echo. With a longer echo time the susceptibility effect of hemosiderin, but also of deoxyhemoglobin and ferritin, increases. This helps to detect microbleeds that are located near veins or in the basal ganglia. These microbleeds can be clearly visible on the first echo, but may be obscured by veins and ferritin-containing structures, such as the basal ganglia, on the second echo. Even large microbleeds, which may be clinically relevant, can be obscured by overlapping structures on the second echo and become visible on the first echo image. Besides that, the first echo image shows the dark microbleeds against a homogeneous, more hyperintense signal of the brain tissue, making the detection of microbleeds easier, as the veins and other structures with high susceptibilities different from hemosiderin are less prominent.

Although the first echo image is helpful to distinguish microbleeds from deoxyhemoglobin in veins and ferritin in the basal ganglia, the distinction between microbleeds and calcifications remains difficult. Calcifications also appear as small hypointense foci on both the first and second echo, mimicking microbleeds. The distinction between microbleeds and calcifications can be made by the characteristic location of calcifications (symmetrical in the basal ganglia, or in the pineal gland or choroid plexus) or by identification of calcifications with the use of computed tomography (CT). A recent study showed an additional possibility to distinguish calcifications on MRI with the use of phase images (27). In this study we used the characteristic location of calcifications, which is in line with other studies investigating microbleeds (3).

The summation of the MIP and the minIP of the first echo image gives an image in which both the arteries and the microbleeds can be visualized. This is useful to analyze the relation between arteries and microbleeds in vivo without use of contrast agents. This is probably a unique feature of dual echo T2*-weighted imaging at ultrahigh field strength. The short echo time needed for the visualization of arteries is field strength-independent, as this depends on the flow velocity in the arteries. At ultrahigh field strength, the susceptibility effect is large enough to visualize microbleeds even at this short echo time. This enables the visualization of both arteries and microbleeds in one image, whereas at lower field strengths the susceptibility effect may not be large enough to visualize the microbleeds at the first echo time. As it is not known whether the etiology of microbleeds involves mainly the arteries or the veins of the brain, combining the information of the first echo, visualizing arteries and microbleeds, and second echo, visualizing veins and microbleeds, in one scan may provide more insight into the etiologic pathway. Furthermore, the detection of microbleeds with the use of the dual echo T2*-weighted scan at 7.0 T MRI can be important for future studies evaluating the prevalence and the prognostic value of microbleeds.

Nowadays, susceptibility-weighted imaging (SWI) is increasingly used for the detection of microbleeds (28–32). In SWI the magnitude and phase images of the T2*-weighted scan are combined to create enhanced contrast between tissues with different susceptibilities (21). However, SWI depends strongly on the voxel aspect ratio, with an optimal voxel aspect ratio of 1:1:4, while we used more isotropic voxels (aspect ratio of 1:1.3:1.7) (33). The additional value of SWI will be minimal for scans with near-isotropic voxels, and we therefore chose T2*-weighted imaging instead of SWI for the detection of microbleeds in our study at 7.0 T.

In conclusion, T2*-weighted imaging using two echo times at 7.0 T combines the advantages of the first and second echo. The first echo utilizes the strong paramagnetic effect of hemosiderin to obtain large contrast between microbleeds and the surrounding tissue on the first echo image, even in the presence of paramagnetic ferritin. The second echo enables visualization of smaller microbleeds than the first echo, as the blooming effect enlarges the observed size of the microbleeds at the second echo image. Furthermore, additional information about the relation between both arteries and microbleeds as well as veins and microbleeds can be obtained from this scan. Visualization of microbleeds and information about their relation with arteries and veins can be useful for future studies investigating the etiology, prevalence, or prognostic value of cerebral microbleeds.

Ancillary