Retrospective bellows-based reconstruction for cardiac MRI: preliminary experience

Introduction: The two most common methods for respiratory compensation of cardiac MR (CMR) studies are breath-holding and navigator-gating. Respiratory bellows is a long-established, readily available method, but has been rarely used in cardiac imaging, with some exceptions [1,2,3]. This study compares the performance of respiratory chest wall and abdomen bellows to navigator-gating (NAV), and the actual superior-inferior displacements of the lung-liver interface and the heart. Using the relationship between bellows and heart-motion, 3D coronary MRI was obtained using a retrospective [4] bellows-based reconstruction. Methods: Acquisition: Five healthy subjects were imaged on a Philips 1.5 T Achieva MR scanner, during free-breathing using a single shot 2D balanced SSFP acquisition to obtain 100 images ECG-gated to diastole acquired in 100 consecutive heart-beats, with NAV and bellows-monitoring. Scan parameters were: 35 cm FOV, 116 x 93 matrix, 8mm slice, TR/TE/θ=2.3ms/1.1ms/50o, sequential acquisition order, 200.7 ms acquisition window. Further, a 3D gradient echo coronary MRI sequence was acquired during free-breathing with NAV and bellows-monitoring with acceptance of all data. Scan parameters were: 27cm FOV, 240 x 244 x13 Nz, TR/TE/θ=5.5ms/1.7ms/30o, 3 mm slices, fat-saturation, T2-prep, ECGgating to diastole. The k-space volume was collected twice, and raw data was exported. Analysis: The superior-inferior displacements of the diaphragm and the heart was compared to the NAV and bellows data. The relationship of the bellows to the lung-liver interface and to the heart was plotted. Using this relationship, criteria for retrospective acceptance/rejection of data were developed and applied for retrospective bellows-based reconstruction of the multi-frame 3D coronary data, with processing in Matlab. Results: Figure 1 shows the relationship between bellows and the lung-liver interface. Table 1 presents the correlations between the bellows signal and superior-inferior displacements measured by the NAV and images. Figure 2 compares a retrospective NAV-based, and bellows-based reconstruction of 3D coronary MRI data, to an image without respiratory compensation.