Validation of a novel technique to extract cine cardiac cycle without respiratory motion from real-time free-breathing images with unsupervised motion correction
Amir Ali Rahsepar1, Haris Saybasili 2, Ahmadreza Ghasemiesfe 1, Bruce Spottiswoode 2, Ann Ragin1, Jeremy Collins 1, and James Carr1

1Radiology, Northwestern University, Chicago, IL, United States, 2Siemens Medical Solutions, Siemens Healthcare, Chicago, IL, United States


In this study, we presented a novel technique to extract cine cardiac cycle without respiratory motion from real-time free-breathing images with unsupervised motion correction.


Breath-held, ECG gated, segmented cine imaging is the reference standard to measure ventricular systolic function and volume with magnetic resonance imaging (MRI). This technique, however, has limitations in patients with arrhythmia and difficulty breath-holding. An alternative strategy, free-breathing real-time (RT) imaging, can acquire each imaging frame in a single step; however, poor temporal/spatial resolution limits the image quality. An unsupervised motion correction technique for free-breathing RT cardiac imaging, reconstructed motion corrected-RT (MOCO)-RT, has been proposed [1] with preliminary validation in healthy volunteers. In this study, we validate this method for clinical use in patients comparing it to the reference standard segmented balanced steady state free precession (bSSFP) cine imaging.


Segmented bSSFP cine images at end expiration and free-breathing RT images (Cartesian, TGRAPPA factor 4) were acquired with the same spatial/temporal resolution (192 base matrix, 45 ms temporal resolution) in 42 patients (age: 54.73±15.25, 71.5% Male) using clinical 1.5-Tesla MR scanners (MAGNETOM Aera and Avanto, Siemens Healthcare, Erlangen, Germany) in this IRB approved study. A full short-axis stack (9-11 slices, 6-8 mm slice thickness) was obtained. RT cardiac images were acquired for approximately 16 beats per slice. For each slice, interpolated images were averaged to obtain an initial reference image corresponding to the most common organ positions for the individual subject, typically a diastolic cardiac phase at end-expiration. Non-rigid unsupervised motion correction was applied, as described elsewhere [1]. Both conventional segmented and MOCO-RT single heartbeat cine images were analyzed to evaluate left ventricular (LV) function, mass and volume (Syngo Argus, Siemens Healthcare, Erlangen, Germany). An experienced radiologist scored images for overall image quality, artifact and noise using a 5-point Likert scale. Wall motion abnormalities (WMA) were scored as a dicohotomous variable (present, absent) in the 16 segment AHA model. These ratings were summed across all 16 segments to derive a global WMA score and scores for basal, mid LV and apical regions. Intraclass correlation coefficient (ICC) was used to assess the reliability of MOCO-RT measurement of LV function and volume using SPSS.


Diagnostic images were acquired in all 42 patients using the MOCO-RT approach. Figure 1 presents images for two subjects with good and poor breath-holding capabilities. ICC showed excellent reliability (ICC>80%) of MOCO-RT with segmented cine in measuring LV function, mass and volume (Table 1). Comparison of the qualitative ratings indicated higher image quality than either segmented cine or conventional RT technique (p<0.05). Artifact ratings were generally lower and noise levels were comparable. Eight subjects with WMA were found in this study. Comparison of WMA scores indicated no differences between MOCO-RT and segmented methods, but RT images showed a superior capability of identifying patients with WMA, particularly in patients with “regional” WMA (Figure 2). However, because of the small number of subjects with WMA the comparison failed to show a statistically significant difference.


We have presented a novel technique that generates single heartbeat, high SNR cine images at end expiration respiratory phase derived from multiple heartbeat, free-breathing RT images. This validation study in clinical patients indicates that LV measurements obtained with MOCO-RT show excellent reliability with the current gold standard segmented cine images. The free-breathing MOCO-RT technique may have considerable clinical utility in cardiac MRI for uncooperative patients who have difficulty with breath holding and in patients with arrhythmia. MOCO-RT technique does not require breath-holding and produces comparable cardiac function, volumes and myocardial mass measurements compared with conventional methods while providing better image quality, less artifact and probably better capability of diagnosing WMA.


No acknowledgement found.


[1] Saybasili Haris, et al. Extracting a cine cardiac cycle without respiratory motion from real-time free-breathing images with unsupervised motion correction, ISMRM 2015.


Figure 1. RT, real-time; MOCO, motion corrected. Figure 1. Short axis images of the heart acquired using real-time, motion corrected real-time and segmented cine techniques in two patients with good breath-holding (A, B, and C) and poor breath-holding ability (D, E, and F).

Figure 2. Comparison of qualitative analysis for diagnosis of wall motion abnormality (WMA).

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)