An area-based imaging biomarker for characterizing coronary artery stenosis with myocardial BOLD MRI

Methods 2D cine SSFP-based BOLD images were acquired in 9 dogs under rest, and adenosine stress with and without LCX stenosis (of varying grades) in a 1.5T scanner. Scan parameters: resolution=1.2x1.2x6mm; flipangle=90; and TR/TE=6.2/3.1ms. Microsphere analysis was used to measure true perfusion. First-pass perfusion and late-enhancement scans were performed to visually confirm perfusion deficits and absence of infarction. Microsphere flow within each AHA segment was summed to obtain total flow per slice. MFR, defined as the ratio of flow between stress and rest was computed. End-systolic (ES) and end-diastolic (ED) images were identified and myocardial borders were traced. Myocardial pixel intensities from rest images were fitted to location-scaled t-distribution to estimate the location (μ) and scale (s) parameters. Affected-Fraction (AF), defined as the ratio of the area of largest contiguous hypointense region (pixel intensity below μ-s) divided by the total area of the myocardium, was computed for both stress (AFSTRESS) and rest (AFREST) cases. Ischemic-Extent (IE), was defined/computed as IE=AFSTRESS/AFREST. For comparison, mean signal intensities of AHA segments corresponding to the LCX territory were normalized by the mean intensity of the entire myocardium to obtain IREST and ISTRESS. Segment-Intensity-Response (SIR), was defined/computed as SIR=ISTRESS/IREST. IE and SIR derived from ES and ED images were each regressed with MFR. ROC analysis was used to examine the diagnostic capacities of IE and SIR metrics to detect critical stenosis at ES and ED.