Time-resolved analysis of coronary vein motion and cross-sectional area

Authors

  • Jonathan D. Suever BS,

    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology / Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • Pierre J. Watson BS,

    1. Department of Radiology & Imaging Sciences, Emory University School of Medicine, Atlanta, Georgia, USA
    Search for more papers by this author
    • Mr. Suever and Mr. Watson contributed equally to this article.

  • Robert L. Eisner PhD,

    1. Department of Radiology & Imaging Sciences, Emory University School of Medicine, Atlanta, Georgia, USA
    Search for more papers by this author
  • Stamatios Lerakis MD,

    1. Department of Cardiology, Emory University School of Medicine, Atlanta, Georgia, USA
    Search for more papers by this author
  • Robert E. O'Donnell MD,

    1. Department of Cardiology, Emory University School of Medicine, Atlanta, Georgia, USA
    Search for more papers by this author
  • John N. Oshinski PhD

    Corresponding author
    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology / Emory University, Atlanta, Georgia, USA
    2. Department of Radiology & Imaging Sciences, Emory University School of Medicine, Atlanta, Georgia, USA
    • Department of Radiology, Emory University School of Medicine, 1364 Clifton Road, Atlanta, GA 30322
    Search for more papers by this author

Abstract

Purpose:

To quantify periods of low motion and cross-sectional area changes of the coronary veins during the cardiac cycle for planning magnetic resonance coronary venograms (MRCV).

Materials and Methods:

Images were acquired from 19 patients with coronary artery disease (CAD) and 13 patients scheduled for cardiac resynchronization therapy (CRT). The displacement and cross-sectional area of the coronary sinus was tracked, and periods of low motion were defined as consecutive time points during which the position of the coronary sinus remained within a 0.67-mm diameter region. Patients were classified as systolic dominant or diastolic dominant based on the relative duration of their low motion periods.

Results:

All CRT patients were classified as systolic dominant, and 32% of these had no separate diastolic rest period. All CAD patients with ejection fraction < 35% were classified as systolic dominant, while all CAD patients with ejection fraction > 35%were diastolic dominant. In 77% of all subjects, the cross-sectional area of the coronary sinus was larger in systole than in diastole.

Conclusion:

The movement of the coronary sinus can be used to classify patients as either having a longer systolic or diastolic rest period. The classification of the CRT patients as systolic dominant suggests that MRCVs be acquired in systole for CRT planning; however, each patient's low motion periods should be categorized to ensure the correct period is being used to minimize motion artifacts. J. Magn. Reson. Imaging 2011;. © 2011 Wiley-Liss, Inc.

CARDIOVASCULAR MAGNETIC RESONANCE imaging is susceptible to artifacts caused by cardiac and respiratory motion. Several methods exist to reduce these artifacts (1). Electrocardiographic (ECG) gating can reduce cardiac motion artifacts by acquiring images over brief acquisition periods during which motions are small and consistent (2). Breathholding can reduce respiratory motion artifacts in two-dimensional acquisitions, but cannot be used when high-resolution, three-dimensional anatomy over a large area is required. Navigator-echo based respiratory gating allows free breathing during acquisition, but at the penalty of lengthening the total acquisition time. In general, these gating methods reduce motion artifacts and improve image quality, but with the penalty of increased scan times.

Whole heart, ECG, and navigator-echo gated, contrast-enhanced MRI techniques have been used to acquire three-dimensional cardiac MR coronary venograms (MRCV) for planning lead placement in cardiac resynchronization therapy (CRT) (3). Visualization of the anatomy and size of coronary veins is important in planning left ventricle (LV) lead placement in CRT (4, 5). The veins move on the order of several vessel diameters during the cardiac cycle. To reduce the effects of cardiac motion during these scans, it is desirable to acquire image data only during periods of low vessel motion, typically assumed to be during mid-diastole (6–8). The temporal location of the low motion period can be used to set the trigger delay in MRCV acquisitions, and the temporal length of the low motion period can be used to set the duration of the acquisition window in MRCV studies. The significant changes in cross-sectional area over the cardiac cycle can also have a large effect on image quality (9, 10). Imaging of the veins at a point in the cardiac cycle when the cross-sectional area is larger may lead to improved visibility.

Previous studies have examined the time-resolved motion and cross-sectional area of the coronary arteries in patients with coronary artery disease (CAD) (11–14). However, only a few limited studies have looked at the motion of the coronary veins (15). A systematic analysis of coronary vein motion in patients scheduled for CRT has not been performed. The purpose of this study was to examine and compare the motion of coronary veins in two groups of patients: patients with CAD and patients scheduled for CRT. The motion of the coronary veins during the cardiac cycle was quantified by identifying the temporal locations and durations of periods of low motion and by analyzing these low motion periods to determine the best image acquisition windows to reduce motion artifacts.

MATERIALS AND METHODS

Patients

Thirty-two patients in two separate groups were studied. Nineteen of the 32 patients (age, 56.1 ± 11.2 years; EF, 25.0–72.7) had coronary artery disease (CAD) with a documented myocardial infarction (MI) at least 6 months before exam. These patients represent those typically included in coronary artery motion studies. Thirteen patients (age, 60.9 ± 12.2 years) were scheduled for CRT, having been classified as New York Heart Association (NYHA) Class III heart failure with QRS duration > 120 ms and ejection fraction (EF) <35%. Written informed consent was obtained from all participants and the protocol was approved by the University's Institutional Review Board (IRB) on human subjects.

MRI Study Protocol

For this study, all subjects underwent a full cardiac MR exam. Analysis was performed on steady-state free procession (SSFP) cine images acquired in the vertical long-axis (two-chamber) orientation with at least 30 frames over the cardiac cycle (R–R interval) on a Siemens Avanto 1.5 Tesla (T) Scanner (Siemens Medical Solutions, Erlangen, Germany) or a Philips Intera 1.5T Scanner (Philips Medical Systems, Best, The Netherlands). Retrospective ECG gating was used in all acquisitions. Acquisition parameters were: acquired matrix size = 192 × 156 to 256 × 195, reconstructed matrix size = 192 × 156 to 256 × 256, field of view (FOV) = 300 × 244 to 418 × 418 mm, flip angle = 65–67°, repetition time (TR) = 2.5–3.4 ms, and echo time (TE) = 1.25–1.7 ms.

Image Review

The coronary sinus was tracked under the assumption that its movement is an indicator of overall coronary vein motion (3, 16, 17). The cross-sectional area and centroid of the coronary sinus were computed for each frame by manual tracing of each vein using in-house software developed in Matlab (The MathWorks, Natick, MA).

To determine low-motion periods, a variant of the Quality Threshold (QT) clustering algorithm was used (18). Clustering was based on the Euclidean distance between the centroid of the coronary sinus computed for all phases of the cardiac cycle. Time points were binned so as to create the largest possible cluster without exceeding a predefined maximum diameter threshold (the quality threshold). To identify a contiguous rest period, the original algorithm was adapted to ensure that only consecutive time points could be clustered. For our analysis, we used a cutoff of 0.67 mm, the pixel size used in our whole-heart coronary vein scan. Using this approach, we were able to identify periods over which the vessel translated less than a pixel in any given direction.

Data Analysis

The temporal location and duration of the periods of low motion were determined as a percent of the cardiac cycle (%CC). The low motion periods were classified as either systolic or diastolic low motion periods based on their temporal location. Systolic low motion periods were defined as any low motion period that began before the onset of left ventricular relaxation. Diastolic low motion periods were defined as any low motion period that began after the smallest left ventricular volume was reached. Each patient was classified into either systolic dominant or diastolic dominant based on the ratio of each patient's systolic to diastolic low motion period duration. Patients with a ratio greater than or equal to 1 (systolic low motion period duration greater than or equal to diastolic low motion period duration) were classified as systolic dominant and those with a ratio less than 1 were classified as diastolic dominant. The temporal location for both the low motion periods was calculated to determine the variability in onset times and duration for the two periods.

The average cross-sectional area of the coronary sinus over the cardiac cycle was determined. The difference in cross-sectional area between systolic low motion and diastolic low motion was compared for each vessel using a paired t-test to determine whether cross-sectional area was significantly larger in either low motion period.

RESULTS

Low Motion Periods

The coronary sinus was well visualized in the two-chamber vertical long-axis cine images in all patients, Figure 1. Nineteen of the 32 patients (59%) were classified as systolic dominant and 13 (41%) were diastolic dominant. All 13 CRT-scheduled patients (100%) were systolic dominant. Thirteen of the 19 (68%) CAD patients were diastolic dominant. Typical displacement curves for the systolic dominant CRT patients and diastolic dominant CAD patients can be seen in Figure 2. In the systolic dominant group, the center of systolic low motion occurred at 46.6 ± 8.2 %CC with a duration of 12.6 ± 5.4 %CC (98.7 ± 41.1 ms). Six of the 19 (32%) systolic dominant patients had no separate diastolic low motion period. Either the diastolic low motion period was completely absent or combined with the systolic low motion period. The diastolic dominant group had a diastolic period of low motion located at 73.1 ± 3.9 %CC with a duration of 20.0 ± 4.9 %CC (168.3 ± 48.7 ms). The systolic low motion occurred at 35.6 ± 5.2 %CC with a duration of 7.6 ± 1.6 %CC (62.6 ± 12.2 ms).

Figure 1.

Vertical long-axis, two-chamber, cine MR images of the heart during systole (a) and diastole (b) showing a cross-section of the coronary sinus (white arrows). The changes in location and cross-sectional area are clearly visible.

Figure 2.

Displacement of the coronary sinus over the cardiac cycle is shown in a typical CRT (a) and CAD patient (b). The low motion periods are denoted by the shaded regions and were defined using consecutive coronary sinus locations that did not exceed the dimensions of a pixel.

All (100%) patients with an EF less than 35% were systolic dominant and all (100%) patients with an EF greater than 35% were diastolic dominant (P < 0.001), Figure 3. The EFs were 21.7±7.8% for the systolic dominant group and 53.6±10.8% for the diastolic group (P < 0.001). The heart rates were 76.9 ± 11.4 beats per minute (bpm) for the systolic group and 72.5 ± 7.3 bpm for the diastolic groups (P = 0.2). Therefore, cardiac cycle length was not significantly different between the two groups.

Figure 3.

EF for the systolic and diastolic dominant patients. The dotted red line shows the 35% EF cutoff value for undergoing CRT. All patients with an EF less than 35% were systolic dominant.

Cross-sectional Area

Six of the 19 systolic dominant patients (32%) had no diastolic low motion period and were therefore excluded from the percent change in cross-sectional area calculations. In 20 of the 26 patients (77%) with both systolic and diastolic rest periods, the cross- sectional area was larger during the systolic low motion period than during the diastolic low motion period (60.1 ± 21.5 mm2 vs. 43.7 ± 22.6 mm2, P < 0.001). Three of the diastolic dominant patients (23%) and 3 of the systolic dominant patients (16%) had a smaller coronary sinus diameter in systole compared with diastole. Of the 13 systolic dominant patients, the area change was 18.6 ± 16.2 % with 3 having larger area during diastole than during systole. The diastolic dominant group showed significantly more change (46.8 ± 29.6 %) in cross-sectional area (P = 0.006) (Figure 4).

Figure 4.

The mean cross-sectional area during systole and diastole analyzed with a paired t-test showing significantly larger area in systole. The red lines indicate the area change for individual patients. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DISCUSSION

This study quantitatively describes the low motion periods of the coronary sinus over the cardiac cycle (Table 1). The major findings of this study are: (i) 100% of CRT-scheduled patients were systolic dominant, (ii) 100% of patients with an EF less than 35% were systolic dominant, (iii) In 32% of the systolic dominant patients, there was no separate diastolic rest period, (iv) In 77% of subjects, the cross-sectional area of the coronary sinus is larger in systole than in diastole, (v) Diastolic dominant patients had greater change in cross-sectional area over the cardiac cycle.

Table 1. Quantitative Analysis of Location, Duration, and Cross-Sectional Area of Low Motion Periods
 Systolic dominantDiastolic dominantP value
  1. LM = low motion; C-SA = cross-sectional area.

Location (%CC)
 Systolic LM46.6 ± 8.235.6 ± 5.2<0.001
 Diastolic LM77.1 ± 5.073.1 ± 3.90.38
Duration (%CC)
 Systolic LM12.6 ± 5.4 (99 ± 41 ms)7.6 ± 1.6 (63 ± 12 ms)0.001
 Diastolic LM6.4 ± 2.3 (55 ± 22 ms)20.0 ± 4.9 (168 ± 49 ms)<0.001
C-SA change (%)18.6 ± 16.246.8 ± 29.60.003
Ejection fraction (%)21.7 ± 7.853.6 ± 10.8<0.001
Heart rate76.9 ± 11.472.5 ± 7.30.20
 SystoleDiastoleP value
C-SA (mm2)59.9 ± 27.634.7 ± 10.0<0.001

Despite previous studies stating that scanning during diastolic low motion is most beneficial for imaging coronary vasculature (8), all of the subjects scheduled for CRT were systolic dominant and therefore scanning them during diastole would result in more motion artifacts. For the systolic dominant patients, beginning a scan during systole instead of during diastole would allow a longer temporal window with low motion and would allow a reduced scan time. However, there was a large variability in systolic duration between patients particularly in the systolic dominant group.

The EF of each patient was observed to be a significant predictor (P < 0.001) of their low motion period classification. All of the CRT scheduled patients had an EF less than 35% as required for this treatment, and all of them were classified as systolic dominant. Additionally, all 6 CAD patients that had an EF less than 35% were also classified as systolic dominant. A patient's EF could potentially be used to determine which low motion period should be used for minimal motion artifacts.

The large variability in the length and position of the low motion periods in this patient population suggests that each patient should be imaged to determine a patient-specific window position and length for MRCV. Of the systolic dominant patients, 32% did not have clearly defined diastolic low motion periods. The subjects without separate diastolic low motion periods typically had very long systolic low motion periods that continued into diastole. These periods were technically both systolic and diastolic low motion because they occurred during both parts of the cardiac cycle but were defined as systolic low motion based on the temporal starting location of the low motion period.

Analysis of the cross-sectional area difference showed that 77% of subjects had a larger area during systole than during diastole presumably due to increased coronary venous pressure. The coronary sinus is the largest of the coronary veins and collects blood from the ventricular veins during systole and empties into the right atrium causing the vessel to contract during systole and dilate during diastole (19). Increased atrial and ventricular pressure in end systole at the end of left ventricular ejection causes dilation of the vessel (10, 20).

When analyzing cross-sectional area between the systolic and diastolic dominant groups, it was observed that the diastolic dominant group showed significantly larger percent changes in area over the cardiac cycle than the systolic dominant group. Six of the systolic dominant subjects had larger cross-sectional area during diastole. Increased cross-sectional areas during systole would improve visualization of the vessel due to its larger size (15).

The clinical importance of this study is that when imaging the coronary veins, one should examine the cine long-axis two-chamber images to determine coronary sinus motion. In patients with low EF, the longer period of low motion will probably be in systole. The combination of larger vessel areas during the systolic low motion periods and the fact that all of the pre-CRT patients had longer systolic low motion periods suggest that systolic imaging would be preferable in obtaining MRCV in pre-CRT patients.

A limitation of this study is that the coronary sinus moves in all three dimensions during the cardiac cycle and therefore the cross-sectional area changes might be due not only to vessel dilation but also to different slice alignments over the cardiac cycle with respect to the vessel. Changes in vessel diameter over the vessel length could be misinterpreted as cross-sectional area changes if the slice plane moves perpendicularly to the vessel. However, previous studies have found that angular deviations below 10 degrees would not cause a significant source of error (21).

In conclusion, the movement of the coronary sinus can be used to classify patients into either systolic dominant or diastolic dominant based on the location of their periods of low motion within the cardiac cycle. All the patients scheduled for CRT and one CAD patient (with an EF < 35%) were classified as systolic dominant, and a period of diastolic low motion was not present in several of these patients. The cross-sectional area was observed to be significantly larger during systole than during diastole leading to improved image resolution of the vessel during systole. The temporal imaging location should be not assumed to be during diastole, and each patient's low motion periods should be categorized before imaging the coronary veins to ensure the correct period is being used to minimize motion artifacts.

Acknowledgements

The authors thank Susan Eder RT(MR) for her help in acquiring the images. J.D.S., received a Graduate Research Fellowship from the National Science Foundation and J.N.O. was funded by the American Heart Association and the National Institutes of Health.

Ancillary