Myocardial BOLD MRI is a non-contrast approach for examining myocardial perfusion. Although recent developments have shown promising technical advancements, current myocardial BOLD MR methods are still limited by: (a) poor spatial coverage and imaging speed; (b) imaging confounders such as heart rate dependency between rest and stress states and coil bias; and (c) imaging artifacts, particularly at 3T. To address these limitations, we developed a heart-rate independent, free-breathing 3D T2 mapping technique at 3T that utilizes near 100% imaging efficiency, which can be completed in 3 minutes with full LV coverage. We tested our method in a canine model with coronary stenosis and validated our findings with simultaneously acquired 13N-ammonia PET perfusion data in a whole-body PET/MR system.Sequence Design: Previous studies have shown that motion-corrected, fast, free-breathing 3D T2 mapping with hybrid trajectory at 3T is possible (1). While this approach minimizes the heart rate dependency of T2 measurements, the use of 2 R-R intervals for signal recovery between segmented acquisitions extends the overall acquisition time and limits its use for whole-heart myocardial BOLD MRI, where data acquisition needs to be performed relatively fast (i.e. within the limited duration of provocative stress (4-6 minutes)). To increase the imaging speed while ensuring robust T2 measurements during provocative stress at 3T, a highly time-efficient and heart-rate independent 3D T2 mapping sequence was developed. To eliminate the heart rate dependency, a saturation recovery (SR) pulse with a constant recovery time was used to eliminate the variability of longitudinal magnetization between readouts. An adiabatic T2 preparation and centric GRE readout with hybrid trajectory was used for highly efficient, off-resonance artifact-reduced, motion-corrected, reconstruction as previously described (1) . The timing diagram for the approach is summarized in Figure 1. Data acquisition: Healthy canines (n=7) and canines with left-anterior-descending (LAD) coronary artery stenosis (n=6) were studied in a clinical PET/MR system (Siemens Medical, Germany). The above described sequence was prescribed in all animals during rest and under adenosine stress (dose: 140 mg/min/kg; TR/TE =3.2/1.6 ms, flip angle = 15°, imaging resolution = 2x2x5 mm3 with 16 partitions and 15% slice over sampling, adiabatic T2 prep pulses and SR recovery time=350ms). Total acquisition time for whole LV coverage was <3 mins. Dynamic 13N-ammonia PET scans were acquired for validation. PET images were analyzed using commercially available qPET software. In healthy dogs, mean myocardial T2 (T2avg) values were measured from basal, mid and apical slices at rest and stress and the corresponding slices were matched to 13N-ammonia PET images to derive the corresponding mean myocardial blood flow (Qavg). Myocardial BOLD response (T2avg (stress):T2avg(rest)) and myocardial perfusion reserve (Qavg (stress):Qavg(rest); MPR) were computed and compared. In animals with LAD stenosis, the perfusion defect regions were identified using an automated algorithm from qPET. The BOLD images were matched to 13N-ammonia PET images to identify the affected territories in the T2 maps. T2avg and Qavg were measured at rest and stress in the affected and remote territories. Myocardial BOLD response and MPR were derived from affected and remote regions and compared against to each other. [RD1]REF? You can just state Yang et al MRM 2015 under references… [RD2]REF? Yang et al MRM 2015 under referencesFigure 2 shows a representative set of BOLD (A) and PET (C) images acquired from a healthy dog under rest and adenosine stress. T2avg measured under adenosine stress (B) were significantly higher than at rest (T2avg: 33.5±1.0 ms (rest) vs. 38.4±3.1 ms (stress), p<0.05)). A similar trend was observed in PET (D) (Qavg: 0.8±0.1 ml/mg/min (rest) vs 2.0±0.9 ml/mg/min (stress); p<0.05). Results from animals with LAD stenosis are shown in Figure 3. A set of PET and BOLD images in an animal with LAD stenosis acquired under adenosine stress are shown (A and C). Perfusion defect was consistently observed in the LAD territory from both PET and BOLD images. Panel B shows myocardial BOLD response was significantly higher in the remote regions (1.09±0.04) compare to the affected regions (1.00±0.03), p<0.05 . A similar trend was also observed with MPR(Remote: 2.8±1.7, Affected: 1.4±1.0, p<0.05; Panel D).The proposed BOLD CMR approach permits rapid whole LV assessment of BOLD changes between rest and adenosine stress. The BOLD responses measured by myocardial T2 were closely correlated with PET perfusion, suggesting that the proposed BOLD CMR method is a viable approach for imaging myocardial perfusion without contrast agents.