Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Apr 2.
Published in final edited form as: Magn Reson Imaging. 2023 Nov 21;105:125–132. doi: 10.1016/j.mri.2023.11.010

On the Loss of Image Contrast in Double-Inversion-Recovery Prepared T2* MRI of Intramyocardial Hemorrhage

Xingmin Guan 1, Xinheng Zhang 1,2, Hsin-Jung Yang 3, Rohan Dharmakumar 1
PMCID: PMC13041346  NIHMSID: NIHMS1947886  PMID: 37993042

Abstract

Purpose:

Studies have shown that double-inversion-recovery (DIR) prepared dark-blood T2*-weighted images result in lower SNR, CNR and diagnostic accuracy for intramyocardial hemorrhage (IMH) detection compared to non-DIR-prepared (bright-blood) T2*-weighted images; however, the mechanism contributing to this observation has not been investigated and explained in detail. This work tests the hypothesis that the loss of SNR on dark-blood cardiac T2*-weighted images of IMH stems from spin-relaxation during the long RF pulses in double inversion preparation, as a result, compromising image contrast for intramyocardial hemorrhage detection.

Methods:

Phantom and in-vivo animal studies were performed to test the hypothesis of the study. An agar phantom was imaged with multi-gradient-echo T2* imaging protocols with and without double-inversion-recovery (DIR) preparation. Image acquisitions were placed at different delay times (TD) after DIR preparation. SNR, T2* and Coefficient of Variation (COV) were measured and compared between DIR-prepared and non-DIR-prepared images. Canines with hemorrhagic myocardial infarctions were scanned at 3.0T with DIR-prepared (dark-blood) and non-DIR-prepared (bright-blood) T2* imaging protocols. DIR-prepared T2* images were acquired with short, medium, and long delay times (TD). SNR, CNR, intramyocardial hemorrhage (IMH) extent, T2* and COV were measured and compared between DIR-prepared T2* images with short, medium, and long delay times (TD) to non-DIR-prepared bright-blood T2* images.

Results:

Phantom studies confirmed the hypothesis that the SNR loss on DIR-prepared T2* images originated from signal loss during DIR preparation. SNR followed T1 recovery curve with increased delay times (TD) indicating that SNR can be recovered with longer time delay between DIR and image acquisition. Myocardial T2* values were not affected by DIR preparation but COV of T2* was elevated. Animal studies supported the hypothesis and showed that DIR-prepared T2* images with insufficient delay time (TD) had impaired sensitivity for IMH detection due to lower SNR and CNR, and higher COV.

Conclusion:

We conclude that lower SNR and CNR on DIR-prepared T2* images originate from signal loss during DIR preparation and insufficient recovery between DIR preparation and image acquisition. Although, the impaired sensitivity can be recovered by extending delay time (TD), it will extend the scan time. Bright-blood T2* imaging protocols should remain the optimal choice for assessment of intramyocardial hemorrhage. DIR-prepared dark-blood T2* imaging protocols should be performed with extra attention on image signal-to-noise ratio when used for intramyocardial hemorrhage detection.

Keywords: Cardiac MRI, Dark-blood, bright-blood, T2*, Intramyocardial hemorrhage

INTRODUCTION

Following reperfusion therapy for acute myocardial infarction, at times blood can extravasate into the interstitial space leading to intramyocardial hemorrhage (IMH) within myocardial infarction (MI). This results in the most severe form of myocardial injury in reperfused MI, which is captured as Stage IV injury in the recently released CCS-AMI classification [1]. Studies have shown that IMH is associated with larger MIs, delayed infarct healing, persistent microvascular obstruction, higher left ventricular volumes, compromised left-ventricular ejection fraction and late-arrhythmogenic risk [2–14]. Therefore, accurate detection of IMH is crucial for identifying hemorrhagic MI patients and in the potential development of therapies against its adverse effects. To date, MRI-based approaches have been used to detect IMH and consequential iron deposition in the convalescent phase of myocardial infarction. These approaches take advantage of the observation that IMH creates local susceptibility-induced distortions in the static magnetic field and accelerates transverse magnetization decay. Notably, T2*-based MRI techniques have emerged as the standard method for detecting hemorrhagic MI as they have high sensitivity for detecting and quantifying the iron content within MI [10, 15].

Cardiac T2* imaging can be performed with either bright-blood or dark-blood. Dark-blood [16] cardiac T2* was developed based on the conventional segmented, multi-gradient-echo bright-blood cardiac T2* technique by adding a double-inversion-recovery (DIR) preparation at R-wave. The DIR preparation consists of one non-selective 180° inversion adiabatic pulse, which is immediately followed by a slice-selective 180° inversion adiabatic pulse to invert the magnetization within the slice of interest. In this scheme, blood within the slice of interest is replaced by the blood with inverted magnetization within left ventricle after R-wave. And readout takes place at the blood-nulling point to ensure that the blood pool appears dark, which allows for clear blood/myocardium interface.

The dark-blood cardiac T2* technique [16] was initially used for assessment of global iron overload diseases [17–19], such as thalassemia [20], since it can suppress blood signal for better delineation of myocardium. More recently, dark-blood technique has become the common choice in the application of cardiac T2* imaging for assessment of myocardial iron overload including local iron overload such as IMH [12, 21, 22].

However, evidence showed that dark-blood technique may not provide equivalent diagnostic accuracy for assessment of myocardial infarctions as bright-blood techniques [23, 24]. In our previous studies [15, 23], we have systematically evaluated the capability of bright-blood and dark-blood cardiac T2* imaging techniques for assessment of IMH. We demonstrated [23] that the conspicuity of IMH is significantly reduced on dark-blood T2* images compared to that observed on bright-blood T2* images, across infarct age in clinical and preclinical settings at 1.5T and 3.0T. Comparing to bright-blood T2* images, dark-blood T2* images showed lower signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and higher Coefficient of Variation (COV), which led to underestimated IMH extent and impaired diagnostic performance. In this work, we further investigated the mechanisms of the previous findings.

In dark-blood techniques, the DIR preparation uses adiabatic inversion pulses to minimize sensitivity to B1 inhomogeneity. However, adiabatic RF pulses are much longer (10 – 20 ms, depending on static field strength) than conventional RF pulses, which can have a non-negligible impact of relaxation [25] on magnetization during the excitation process given the short T2 of myocardial tissue (T2 of myocardial tissue is about 45 ms at 3.0T) [26]. Particularly the impact is doubled by the use of two consecutive adiabatic RF pulses for magnetization preparation in DIR-prepared dark-blood techniques. Accordingly, we hypothesize that the loss of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) on dark-blood cardiac T2* images of IMH stems from signal loss [25, 27] during the double inversion preparation. To test our hypothesis, we performed phantom studies and validated our results with an animal model.

Methods

All images were acquired in a 3T MRI system (Verio, Siemens Healthcare, Erlangen, Germany). Phantom images were acquired using a 12-channel head coil; and in-vivo images were acquired with 32-channel phased-array coil.

Phantom Study

A phantom study was performed for verification of magnetization loss due to double-inversion-recovery preparation and to test whether altering the delay time (TD) between DIR and image acquisition can improve signal-to-noise ratio (SNR). Phantoms consisted of two plastic tubes (Tube I and II), both with 2% agar but one (Tube I) was mixed with 0.5 mM iron oxide (Ferumoxytol, Feraheme, AMAG Pharmaceuticals, United States) to arrive at solutions with different relaxation rates.

Phantoms were immersed in water and imaged at 3.0T MR system (Verio, Siemens Healthcare, Erlangen, Germany) with simulated ECG signals, to be consistent with in-vivo studies, at a heart rate of 67 beats per minute. Multi-echo T2*-weighted images with and without DIR preparation were acquired. To evaluate the effect of delay time (TD, Figure 1) between DIR preparation and readout, DIR-prepared images were acquired with TD = 0, 100, 200, 300, 400, 500, 600, 700, 860, 1060, 1260, 1460, 1660 ms between DIR preparation and readout. DIR prepared images with TD < 700 ms were acquired with readout at every heartbeat; and DIR prepared images with TD > 700 ms were acquired with readout at every other heartbeat. T2*-weighted images without DIR preparation were acquired for reference.

Figure 1.

Figure 1.

Open in a new tab

Timing diagram for double-inversion recovery (DIR) preparation.

Detailed scanning parameters are as follows. T1 modified Look-Locker inversion recovery (T1-MOLLI) was used to determine T1 relaxation rate of the phantom with 8 inversion times (TI) with 2 Look-Locker cycles, TE = 1.22ms, minimum TI = 120 ms, TI increment = 80 ms, flip angle = 35°, bandwidth = 1085 Hz/Px, spatial resolution = 1.2×1.2×6.0 mm3. All T2* images were acquired with segmented, multi-gradient-echo readout with 8 echoes, TE = 1.58 ms − 10.89 ms, ΔTE = 1.33 ms, TR = R-R interval, number of segments = 7, flip angle = 18°, bandwidth = 925 Hz/pixel, spatial resolution = 1.1×1.1×6.0 mm3, GRAPPA acceleration factor=2, and 200% of dark-blood slice thickness was used for the second inversion pulse of DIR preparation.

Animal Study

Consistent with the protocol approved by the Institutional Animal Care and Use Committee (IACUC), hemorrhagic MIs were created in canines (n = 9) by occluding the left-anterior descending coronary artery (LAD) for 3 hours, followed by reperfusion as described previously [23, 28]. Animals were allowed to recover for 2 months. Subsequently, animals were intubated and anesthetized with isoflurane (1–1.5 %/volume) during cardiac MR scans. After localizer, short axis slice-matched, breath-held, ECG triggered, segmented, multi-echo bright-blood T2*-weighted and DIR-prepared dark-blood T2*-weighted images were acquired. Late-gadolinium enhancement (LGE) images were acquired 10 minutes post-injection of 0.15 mmol/kg gadolinium contrast agent (Magnevist; Bayer AG, Berlin, Germany) as reference of MIs.

All bright-blood and dark-blood T2*-weighted images were acquired at diastole. For dark-blood image acquisition, DIR preparation was placed at different times to adjust delay time (TD) between DIR preparation and readout. Dark-blood T2*-weighted images with short, medium, and long delay times (TD) between DIR preparation and readout were acquired. Due to variation of heart rate in animals, TD was adjusted between 500 to 700 ms for medium TD group, which is also the standard acquisition for dark-blood cardiac T2* images in the clinical setting. Timing of DIR preparation was adjusted to match the delay time (TD) required for short (TD = 280 ms) and long (TD = 1200 ms) TD groups. Images with short and medium TD were acquired with readout performed at every heartbeat. Due to limited R-R intervals, images with long TD were acquired by performing readout every other heartbeat at diastole and DIR preparation during the previous heartbeat.

Detailed scanning parameters are as follows. All T2* images were acquired with segmented, multi-gradient-echo readout with 6 echoes, TE = 2.32 ms – 13.92 ms, ΔTE = 2.32 ms, TR = R-R interval, number of segments = 7, flip angle = 12°, bandwidth = 1185 Hz/pixel, spatial resolution = 1.6×1.6×6.0 mm3, GRAPPA acceleration factor=2, and 200% of dark-blood slice thickness was used for the second inversion pulse of DIR preparation. LGE images were acquired using a gradient-echo sequence with optimal TI for nulling the remote myocardium, TR/TE = 1 R-R interval/1.48 ms, flip angle = 20°, bandwidth = 585 Hz/pixel at a spatial resolution of 1.3×1.3×6.0 mm3.

Image Analysis

All image analyses were performed with CVI42 (Circle Cardiovascular Imaging, Calgary, Alberta, Canada). Remote myocardium was identified as the region absent of hyperintensity on LGE images. MI zone was defined as the region with mean signal intensity (SI) of at least 5 standard deviations (SD) greater than that of a reference region of interest (ROI) drawn in remote myocardium. IMH was identified using Mean – 2SD approach [10, 23, 29]. MI zones were identified to be hemorrhagic if there were hypointense cores within MI on T2*-weighted images (TE = 13.92 ms) with a mean signal intensity 2 SD lower than that of the reference region in the remote myocardium. Subsequently, territories positive for IMH were copied onto T2* maps for determination of T2*.

The area of IMH on each slice measured from T2*-weighted images were normalized by the area of MI measured from the corresponding slice-matched LGE images and reported as IMH Extent. SNR and CNR values were computed from T2*-weighted images. All SNR and CNR measured from DIR-prepared dark-blood (DB) T2*-weighed images (TE = 13.92 ms) were normalized by SNR and CNR measured from bright-blood (BB) T2*-weighted images (TE = 13.92 ms) respectively and reported as relative SNR and relative CNR:

RelativeSNR=SIremote−DB/σair‐DBSIremote−BB/σair−BB, #(1)
RelativeCNR=SIremote−DB−SIIMH−DB/σair‐DBSIremote−BB−SIIMH−BB/σair−BB, #(2)

Where SIremote is the mean intensity of remote myocardium, SIIMH is mean intensity of IMH region and σair is the SD of signal intensity of background air, DB is short for dark-blood and BB is short for bright-blood.

Coefficient of variation was evaluated in myocardial regions remote from infarcts on T2*-weighted images (TE = 13.92 ms) as standard deviation of signal intensity σSI over mean signal intensity (SI) of the region of interest.

COV=σSISI #(3)

Statistical Analysis

Statistical analysis was performed using IBM SPSS Statistics 23 (IBM Corp., Armonk, New York). Normality of continuous data was determined by using the Shapiro-Wilk test and quantile-quantile plots. Normally distributed variables were compared using repeated measures ANOVA. Repeated measures from each heart were nested for analysis. Pairwise comparisons for non-normal data were performed using the Mann-Whitney U test, and paired t-test for normal data.

Results

Phantom Studies

Representative T2*-weighted images and T2* maps

Figure 2 shows our findings from phantom studies. In Figure 2A, the first two rows are showing T2*-weighted images of the agar phantom (Tube I and Tube II) with and without DIR preparation at the same window level. Column 1 shows T2*-weighted images acquired without DIR-preparation. Column 2 – 14 show T2*-weighted images acquired with DIR-preparation with increasing delay time (TD) from 0 ms (column 2) to 1660 ms (column 14) between DIR-preparation and image acquisition. Corresponding T2* maps of Tube I and Tube II are shown in the last two rows in Figure 2A with colormap.

Figure 2. Phantom Studies: Tube #I, 2% agar with 0.5mM iron oxide (T1 = 840.3 ms); Tube #II, 2% agar (T1 = 1834.2 ms).

Figure 2.

Open in a new tab

A. T2*-weighted images and T2* maps obtained from cross sectional views of phantoms without DIR preparation and with DIR preparation followed by increasing delay time (TD) prior to readout. Delay Time (TD) = 0 (2), 100 (3), 200 (4), 300 (5), 400 (6), 500 (7), 600 (8), 700 (9), 860 (10), 1060 (11), 1260 (12), 1460 (13), 1660 (14) ms. B. Relative SNR of DIR prepared images. Dash lines are simulated T1 relaxation curves determined based on T1 values of phantoms. C. T2* from DIR prepared and non-DIR-prepared scans. Scatters show T2* of DIR-prepared images with different TDs and dash line shows T2* of non-DIR-prepared results. SD denotes standard deviation. D. COV of T2*-weighted images. Scatters show COV of DIR-prepared images with different TDs and dash line shows COV of non-DIR-prepared results. E. COV of T2*-maps. Scatters show COV of DIR-prepared T2* maps with different TDs and dash line shows COV of non-DIR-prepared results.

Relative SNR

In Figure 2A, all T2*-weighted images were windowed and leveled to the same level. Visible signal loss was observed on DIR prepared images in both tubes (e.g., column 2 vs. column 1 in Figure 2A first two rows). SNR of DIR-prepared T2*-weighted images with different TD of each tube were measured and normalized by SNR of non-DIR prepared T2*-weighted image and reported as relative SNR in Figure 2B (square markers with whiskers). When TD = 0 ms (corresponding to column 2 in Figure 2A first two rows), which means image was acquired immediately after DIR preparation, relative SNR between DIR-prepared T2*-weighted image and non-DIR prepared T2*-weighted image is 0.622 for Tube I and 0.620 for Tube II, indicating a nearly 40% of SNR loss. With increasing delay time (TD) between DIR preparation and image acquisition, relative SNR is recovered to 0.959 for Tube I and 0.904 for Tube II at TD = 1660 ms, corresponding to column 14 in Figure 2A first two rows. T1 relaxation rate of Tube I and Tube II were measured by T1-MOLLI. T1 of Tube I is 840.3 ± 27.0 ms and of Tube II is 1834.2 ± 35.1 ms. T1-relaxation recovery simulations with initial magnetization (M0) of 0.622, T1 relaxation rate of 840 ms for Tube I, and initial magnetization of 0.620, T1 relaxation rate of 1834 ms for Tube II were performed and shown as dash lines in Figure 2B. For both tubes, recovering SNR with increasing TDs showed similar trend with T1 relaxation curve.

T2* Measurements

Figure 2C shows results of T2* measurements in phantom studies. T2* values measured from DIR-prepared T2* maps with different TDs were shown as square markers with whiskers showing standard deviation. As reference, T2* value measured from non-DIR T2* map is shown as a horizontal dash line. The surrounding color-filled boxes (light blue for Tube I, light red for Tube II) represent standard deviation of T2* of non-DIR preparation. T2* measured on non-DIR prepared map is 25.8 ± 1.6 ms for Tube I and 38.9 ± 4.2 ms for Tube II. No significant differences were found between non-DIR-prepared and DIR-prepared groups with respect to T2* values.

COV

Coefficient of variation (COV) of T2*-weighted images and T2* maps acquired with various delay times (TD) are shown in Figure 2D and 2E, respectively. The scatter plots connected by straight lines represent COV measured from DIR-prepared images with increasing delay times (TD). Dash lines represent COV measured from non-DIR prepared images. In general, DIR prepared T2*-weighted images showed higher COV compared to non-DIR prepared T2*-weighted images for both tubes. However, a decreasing trend of COV were observed with increasing delay times (TD). In Figure 2E, for Tube I (blue), all DIR prepared T2* maps showed higher level of COV compared to non-DIR prepared T2* maps; and for Tube II (red), DIR prepared T2* maps with delay time (TD) less than 1260ms (except for TD = 1060ms) showed higher COV compared to non-DIR prepared T2* maps.

In-vivo Canine Studies

Representative T2*-weighted images and T2* maps

Representative T2*-weighted (TE = 13.92 ms) and LGE images acquired from a canine with hemorrhagic MI are shown in Figure 3. Bright blood T2*-weighted images were acquired with no-DIR preparation. Dark-blood T2*-weighted images were acquired with DIR preparation with short, medium and long TD between DIR and readout. Raw images are shown with epi- and endocardium contours. IMH and MI territories are highlighted on T2*-weighted and LGE images respectively and shown as processed images. Corresponding T2* maps with varying TDs were generated by pixel-wise fitting and shown at the bottom row in Figure 3.

Figure 3. Representative findings from Canine Studies:

Figure 3.

Open in a new tab

Representative T2*-weighted images and corresponding T2* maps obtained from a canine with hemorrhagic MI without DIR and with DIR with different TDs. Raw images are shown with epi- (green line) and endocardium (red line) contours. Intramyocardial hemorrhage (IMH) identified using Mean-2SD criterion is highlighted in yellow shade (Processed, middle row). Raw and processed late gadolinium enhancement (LGE) images are shown for reference. Myocardial infarction (MI) identified on LGE images by mean+5SD criterion is highlighted in blue shade (Processed, middle row).

Relative SNR and CNR

Figure 4 shows relative SNR (Figure 4A) and relative CNR (Figure 4B) measured from DIR-prepared dark-blood T2*-weighted images normalized to bright-blood T2*-weighted images without DIR preparation. Compared to non-DIR prepared bright-blood T2*-weighted images, SNR of DIR-prepared T2*-weighted images was significantly reduced with short and medium TDs. On average, SNR of DIR prepared T2*-weighted images was reduced by 35.7 ± 21.4% (short TD, p<0.01) and 27.3 ± 14.8% (medium TD, p<0.01) compared to bright-blood T2* images. No significant difference in SNR was found between groups without DIR preparation and with long TD DIR preparation (0.5 ± 20.0%, p = 0.76). Similar results were found on CNR analysis. Compared to that of bright-blood T2* images acquired without DIR preparation, CNR was significantly reduced by 34.1 ± 26.2% (p<0.01) and 29.5 ± 17.2% (p<0.01) in short and medium TD DIR-prepared images, respectively. CNR of long TD DIR-prepared dark-blood T2*-weighted images were not different (6.5 ± 21.6%, p = 0.92) from bright blood T2*-weighted images acquired without DIR preparation.

Figure 4. Canine Studies - Effect of Time Delay (TD) on Relative SNR and CNR. A.

Figure 4.

Open in a new tab

Relative SNR of DIR-prepared T2*-weighted images (TE = 13.92 ms) with short, medium, and long delay times (TD) normalized by SNR from non-DIR prepared T2* images. B. Relative CNR of DIR-prepared T2*-weighted images (TE = 13.92 ms) with short, medium, and long delay times (TD) normalized by CNR from non-DIR prepared T2* images. CNR was measured between regions of intramyocardial hemorrhage (IMH) and remote myocardium. Significant differences of relative SNR and CNR were found between non-DIR prepared and DIR prepared groups with short and medium TDs. (** p<0.01, NS: Not Significant)

Coefficient of Variation (COV)

The COV measured from T2*-weighted images and T2* maps from animal studies are shown in Figure 5A and 5B, respectively. In remote regions of myocardium on T2*-weighted images, COV was elevated by 57.4 ± 40.0 % (short TD, p<0.01) and 36.9 ± 34.2 % (medium TD, p<0.01) with DIR preparation compared to that without DIR preparation, but not different from COV with long TD (7.61 ± 30.0 %, p = 0.41). COV of T2* measured from T2* maps showed similar trend: COV of DIR-prepared T2* in the remote myocardium were elevated by 90.1 ± 69.2 % (short TD, p<0.01) and 66.4 ± 61.0 % (medium TD, p<0.01) compared to COV of non-DIR prepared T2* maps; but not significantly different from COV with long TD (7.68 ± 46.3 %, p = 0.88).

Figure 5. Canine Studies - Coefficient of Variation measured from T2*-weighted images (A) and T2*-map (B).

Figure 5.

Open in a new tab

COV was significantly elevated on both T2*-weighted and T2* maps with DIR-preparation with short and medium TDs compared to those on non-DIR prepared images. No significant differences were found between COV from DIR-prepared T2*-weighted and T2* maps with long TD compared to non-DIR-prepared T2*-weighed and T2* maps respectively. (** p<0.01, NS: Not Significant)

IMH Extent

IMH extent was measured using Mean-2SD criterion on bright-blood T2*-weighted images without DIR preparation and dark-blood T2*-weighted images with DIR-preparation with various TDs. Results are shown in Figure 6. On average, IMH extent was significantly underestimated on DIR-prepared T2*-weighted images with short (by 40.4 ± 19.0%, p<0.01) and medium (by 29.9 ± 19.6%, p<0.05) TD, respectively. Difference between IMH extent measured from long TD DIR-prepared dark-blood T2*-weighted images and non-DIR-prepared bright-blood T2*-weighted images is 4.8 ± 8.4 % (p = 0.07). No significant difference was found between groups of DIR-prepared images with long TD and without DIR preparation.

Figure 6. Canine Studies – IMH Extent measured from T2*-weighted images with and without DIR preparation.

Figure 6.

Open in a new tab

IMH was significantly underestimated with short or medium delay TD on DIR-prepared dark-blood T2*-weighted images relative to no DIR preparation (* p<0.05, ** p<0.01, NS: Not Significant).

T2* Measurements

T2* values of IMH and remote myocardium measured from DIR-prepared dark-blood T2* maps and non-DIR-prepared bright-blood T2* maps are listed in Table 1. No difference was found between T2* values of remote myocardium measured from T2* maps constructed from T2*-weighted images acquired with and without DIR preparation. T2* values of IMH regions determined using Mean-2SD criterion on DIR-prepared images with short TD was 16.2 ± 1.9 ms, which was significantly lower than that measured from images acquired without DIR preparation (T2* = 18.7 ± 2.4 ms, p<0.05). Similar observation was found on T2* values of IMH determined from DIR-prepared images with medium TD, which was 17.2 ± 2.9 ms (p<0.05). T2* of IMH determined using Mean-2SD criterion on long TD DIR-prepared T2* maps (T2* = 18.7 ± 3.0 ms, p = 0.14) were not different from T2* measured from T2*maps acquired without DIR preparation.

Table 1.

T2* of Remote and Chronic IMH Territories Identified using Mean-2SD Criterion.

T2* (ms) DIR-prepared No DIR Preparation
Short TD Medium TD Long TD
Remote Myocardium 28.7 ± 2.6 28.1 ± 1.7 27.8 ± 2.4 27.7 ± 2.1
IMH (Mean-2SD) 16.2 ± 1.9 * 17.2 ± 2.9 * 18.7 ± 3.0 18.7 ± 2.4
Open in a new tab
*

(p<0.05)

Discussion

In this work, we tested whether the observed SNR and CNR losses in the identification of IMH on DIR-prepared dark-blood T2* images, compared to the T2* images acquired without dark-blood preparation (i.e., bright-blood images), originated from the signal losses associated with double-inversion recovery preparation. We studied this through a series of studies in phantoms and using an animal model of hemorrhagic MI.

Phantom studies confirmed signal losses from the double-inversion-recovery preparation. The lowest SNR was found in T2* images with image acquisition immediately after DIR preparation (TD = 0 ms) and SNR recovered with increased delay time (TD = 100 to 1660 ms) between DIR preparation and image acquisition. The mitigated signal loss along T1 recovery curve with longer delay time (TD) ruled out possible sources of signal loss from image acquisition side but leaves the only source from the DIR preparation which verified our hypothesis. The deviation from T1 recovery curve after TD of 860ms on Figure 2B is due to the extra recover time after readout when applying readout every 2 heartbeat instead of 1. In-vivo findings were consistent with phantom studies.

Compared to that of non-DIR-prepared bright-blood T2*-weighted images, lower SNR of dark-blood T2*-weighted images with insufficient magnetization recovery from DIR preparation with short and medium TD were found. SNR was recovered to equal level as that of bright-blood T2* images with sufficient delay time (long TD) between DIR and readout.

The loss of SNR has affected image contrast between remote myocardium and hemorrhagic territories (Figure 4B), therefore, led to compromised capability of diagnosis of IMH extent. Furthermore, higher noise level (Figure 5) can aggravate the situation. Since the reduction of IMH extent measured from insufficient recovered DIR-prepared dark-blood T2*-weighted images were significant, it likely impairs the sensitivity for detecting mild iron remnants from IMH. Our findings of reduced SNR and CNR on T2*-weighted images, which lead to reduction in IMH extent measured using Mean-2SD method is in line with results in chronic phase of MIs at 3.0T from previous work [23]. When SNR and CNR recovered to the same level as those on T2*-weighted images acquired without DIR preparation, the underestimation of IMH extent was marginalized.

In addition to reduced SNR and CNR, COV was also found significantly elevated compared to that of non-DIR-prepared T2*-weighted images. The increase in COV is expected to further impair the sensitivity for detecting iron remnants from IMH, particularly when using mean-2SD approach for quantifying IMH-related mild iron depositions on T2*-based images.

Even though SNR, CNR, COV and diagnostic accuracy were recovered with sufficient recover time after DIR, there are remaining issues with this remedy. For example, extending delay time (TD) between DIR preparation and image acquisition will result in missing the blood nulling point (Figure 3-Long TD), which compromises the advantage of dark-blood imaging techniques. In addition, extending delay time (TD) may result in image acquisition in the next cardiac cycle and elongate total scanning and breath holding time, which is unfavorite in clinical applications. Also, details of the signal loss during DIR preparation such as spin-relaxation and inversion efficiency need to be investigated in future works.

In this work, we also investigated the influence of DIR preparation on T2* maps. T2* measured from the phantoms showed no significant difference in T2* values between groups with and without DIR preparation, supporting the notion that the signal losses induced by DIR did not affect T2* decay. Animal studies showed that DIR preparation did not affect T2* values of remote myocardium on T2* maps which is consistent with phantom studies. Compared to T2* values within IMH territories measured from non-DIR prepared T2* maps, T2* values of IMH measured from DIR-prepared T2* maps with short and medium TDs by mean-2SD criterion are significantly lower, which may be explained by the heterogenous distribution of iron deposits within the infarction zone.

Both phantom and animal studies showed that DIR preparation did not affect T2* values on T2* maps, T2* maps based on DIR preparation may be an alternative solution to compensate for signal loss during DIR preparation on T2*-weighted images, however this approach should still be considered with caution. First, DIR induced signal loss on T2*-weighted images can impair the goodness of T2* fitting. This is evidenced by the blue and black areas on T2* maps with DIR-preparation at short and medium TDs (see Figure 3) indicating the failure of T2* fitting with r < 0.5. In some cases, when SNR is reduced below critical thresholds at long echo times, it can introduce unexpected bias in T2* analysis. Second, elevated COV on T2* maps with medium TD DIR preparation, which is the standard dark-blood DIR application scheme, suggests higher heterogeneity of myocardium which may affect diagnostic accuracy for IMH detection. Third, T2* maps are significantly affected by the off-resonance artifacts from the heart-lung interface compared to T2*-weighted images (See Figure 3), which makes it nearly impossible to discriminate between regions affected by off-resonance and non-hemorrhagic MI territories when using T2* threshold for T2* analysis.

Conclusion

The mechanism contributing to lower SNR and CNR on dark-blood T2* images was shown to result from signal loss during double-inversion-recovery preparation and it can be mitigated by imposing a significantly longer recovery time between DIR and readout. However, this has the added disadvantage of missing the blood-nulling point, ultimately compromising the value of dark-blood T2*-based approach. It can also increase scan time which is not desirable in clinical settings for imaging of patients with IMH. IMH assessment based on T2* maps can be used as an alternative to T2*-weighted images when DIR-prepared dark-blood T2* imaging technique is used; however, T2* maps will also suffer from higher heterogeneity with higher COV and lower SNR levels, goodness of T2* fitting and off-resonance artifacts should be carefully considered.

Acknowledgement

This work was supported in part by NIH/NHLBI HL133407, HL136578 and HL147133.

Footnotes

Statement

This is to certify that all authors have seen and approved the final version of the manuscript. This manuscript is the author’s original work, is not under consideration anywhere else and none of its contents have been previously published. Collectively we have no conflicts of interest to report that pertain to this work.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  • 1.Kumar A, et al. , The Canadian Cardiovascular Society Classification of Acute Atherothrombotic Myocardial Infarction Based on Stages of Tissue Injury Severity: An Expert Consensus Statement. Canadian Journal of Cardiology, 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Liu T, et al. , Intramyocardial Hemorrhage and the “Wave Front” of Reperfusion Injury Compromising Myocardial Salvage. Journal of the American College of Cardiology, 2022. 79(1): p. 35–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Kloner RA, Does reperfusion injury exist in humans? Journal of the American College of Cardiology, 1993. 21(2): p. 537–545. [DOI] [PubMed] [Google Scholar]
  • 4.Fishbein MC, et al. , The relationship of vascular injury and myocardial hemorrhage to necrosis after reperfusion. Circulation, 1980. 62(6): p. 1274–1279. [DOI] [PubMed] [Google Scholar]
  • 5.Olafsson B, et al. , Reduction of reperfusion injury in the canine preparation by intracoronary adenosine: importance of the endothelium and the no-reflow phenomenon. Circulation, 1987. 76(5): p. 1135–1145. [DOI] [PubMed] [Google Scholar]
  • 6.Betgem RP, et al. , Intramyocardial haemorrhage after acute myocardial infarction. Nature Reviews Cardiology, 2015. 12(3): p. 156. [DOI] [PubMed] [Google Scholar]
  • 7.Roberts CS, Schoen FJ, and Kloner RA, Effect of coronary reperfusion on myocardial hemorrhage and infarct healing. The American journal of cardiology, 1983. 52(5): p. 610–614. [DOI] [PubMed] [Google Scholar]
  • 8.Lie J, et al. , Hemorrhagic myocardial infarction associated with aortocoronary bypass revascularization. American heart journal, 1978. 96(3): p. 295–302. [DOI] [PubMed] [Google Scholar]
  • 9.Ganame J, et al. , Impact of myocardial haemorrhage on left ventricular function and remodelling in patients with reperfused acute myocardial infarction. European heart journal, 2009. 30(12): p. 1440–1449. [DOI] [PubMed] [Google Scholar]
  • 10.Kali A, et al. , Chronic manifestation of postreperfusion intramyocardial hemorrhage as regional iron deposition: a cardiovascular magnetic resonance study with ex vivo validation. Circ Cardiovasc Imaging, 2013. 6(2): p. 218–28. [DOI] [PubMed] [Google Scholar]
  • 11.Eitel I, et al. , Prognostic value and determinants of a hypointense infarct core in T2-weighted cardiac magnetic resonance in acute reperfused ST-elevation-myocardial infarction. Circ Cardiovasc Imaging, 2011. 4(4): p. 354–62. [DOI] [PubMed] [Google Scholar]
  • 12.Mather AN, et al. , Reperfusion haemorrhage as determined by cardiovascular MRI is a predictor of adverse left ventricular remodelling and markers of late arrhythmic risk. Heart, 2011. 97(6): p. 453–459. [DOI] [PubMed] [Google Scholar]
  • 13.Cokic I, et al. , Iron-Sensitive Cardiac Magnetic Resonance Imaging for Prediction of Ventricular Arrhythmia Risk in Patients With Chronic Myocardial Infarction: Early Evidence. Circ Cardiovasc Imaging, 2015. 8(8). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Behrouzi B, et al. , Action of iron chelator on intramyocardial hemorrhage and cardiac remodeling following acute myocardial infarction. Basic Research in Cardiology, 2020. 115(3): p. 1–18. [DOI] [PubMed] [Google Scholar]
  • 15.Avinash Kali RT, Dharmakumar Rohan, Detection of Acute Reperfusion Myocardial Hemorrhage with Cardiac MR Imaging: T2 versus T2*. Radiology, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.He T, et al. , Black-blood T2* technique for myocardial iron measurement in thalassemia. J Magn Reson Imaging, 2007. 25(6): p. 1205–9. [DOI] [PubMed] [Google Scholar]
  • 17.Anderson L, Cardiovascular T2-star (T2*) magnetic resonance for the early diagnosis of myocardial iron overload. European Heart Journal, 2001. 22(23): p. 2171–2179. [DOI] [PubMed] [Google Scholar]
  • 18.Westwood MA, et al. , Normalized left ventricular volumes and function in thalassemia major patients with normal myocardial iron. Journal of Magnetic Resonance Imaging: An Official Journal of the International Society for Magnetic Resonance in Medicine, 2007. 25(6): p. 1147–1151. [DOI] [PubMed] [Google Scholar]
  • 19.Kondur AK, et al. , Quantification of myocardial iron overload by cardiovascular magnetic resonance imaging T2* and review of the literature. Clinical Cardiology: An International Indexed and Peer- Reviewed Journal for Advances in the Treatment of Cardiovascular Disease, 2009. 32(6): p. E54–E58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Ali S, et al. , Current status of beta-thalassemia and its treatment strategies. Molecular Genetics & Genomic Medicine, 2021. 9(12): p. e1788. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Kidambi A, et al. , The effect of microvascular obstruction and intramyocardial hemorrhage on contractile recovery in reperfused myocardial infarction: insights from cardiovascular magnetic resonance. Journal of Cardiovascular Magnetic Resonance, 2013. 15(1): p. 58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.O’Regan DP, et al. , Reperfusion hemorrhage following acute myocardial infarction: assessment with T2* mapping and effect on measuring the area at risk. Radiology, 2009. 250(3): p. 916–922. [DOI] [PubMed] [Google Scholar]
  • 23.Guan X, et al. , Assessment of intramyocardial hemorrhage with dark-blood T2*-weighted cardiovascular magnetic resonance. Journal of Cardiovascular Magnetic Resonance, 2021. 23(1): p. 1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Payne AR, et al. , Bright-blood T2-weighted MRI has higher diagnostic accuracy than dark-blood short tau inversion recovery MRI for detection of acute myocardial infarction and for assessment of the ischemic area at risk and myocardial salvage. Circ Cardiovasc Imaging, 2011. 4(3): p. 210–9. [DOI] [PubMed] [Google Scholar]
  • 25.Norris DG, Lüdemann H, and Leibfritz D, An analysis of the effects of short T2 values on the hyperbolic-secant pulse. Journal of Magnetic Resonance (1969), 1991. 92(1): p. 94–101. [Google Scholar]
  • 26.Wang G, et al., Influence of Myocardial Hemorrhage on Staging of Reperfused Myocardial Infarctions With T2 Cardiac Magnetic Resonance Imaging. JACC: Cardiovascular Imaging, 2019. 12(4): p. 693–703. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Kellman P, Herzka DA, and Hansen MS, Adiabatic inversion pulses for myocardial T1 mapping. Magnetic resonance in medicine, 2014. 71(4): p. 1428–1434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yang HJ, et al. , Free-breathing, motion-corrected, highly efficient whole heart T2 mapping at 3T with hybrid radial-cartesian trajectory. Magn Reson Med, 2016. 75(1): p. 126–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Chen Y, et al. , Quantification of myocardial hemorrhage using T2* cardiovascular magnetic resonance at 1.5T with ex-vivo validation. Journal of Cardiovascular Magnetic Resonance, 2021. 23(1): p. 104. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES