Dynamic Fractional Myocardial Blood Volume Mapping Using Ferumoxytol-enhanced CMR Multitasking with Subspace Dose Harmonization

ABSTRACT Background Fractional myocardial blood volume (fMBV) represents the vascular compartment within myocardial tissue and may reflect changes in microvascular status and health. Conventionally derived from multi-dose Modified Look-Locker Inversion (MOLLI) recovery T1 mapping with ferumoxytol, MOLLI-based fMBV requires repeated breath-holds, inter-dose image registration, and static end-diastolic estimates. CMR Multitasking enables free-breathing, non-electrocardiogram (ECG)-gated imaging and offers a subspace that captures shared motion and contrast dynamics. This subspace representation can be exploited to harmonize multi-dose acquisitions directly in parameter space, enabling motion-resolved dynamic fMBV mapping while reducing reliance on post-hoc image-space registration. Methods Thirteen healthy volunteers (11 females, 28.2±10.0 years) and twelve ischemic heart disease (IHD) patients (1 female, 67.2±7.1 years) underwent multi-dose ferumoxytol imaging at 3.0 T using both breath-held ECG-triggered 5(3)3(3)3 MOLLI and free-running 2-dimensional Multitasking T1 mapping sequences. MOLLI-Multitasking agreement was assessed in healthy volunteers; patient data were used for exploratory dynamic fMBV analysis. Multitasking data were harmonized across respiratory- and cardiac-motion subspace dimensions for cross-dose alignment. Pixel-wise fMBV was estimated using a two-compartment water exchange model. fMBV agreement with MOLLI was assessed using Bland–Altman analysis. Dynamic fMBV changes were also evaluated from the Multitasking data. In IHD patients, fMBV kinetic metrics (upslope and diastolic-to-systolic distribution) for cycle-dependent variation in dynamic fMBV were computed to assess systolic compression and diastolic recoil. Cardiac positron emission tomography (PET) derived perfusion and CMR-derived strain and late gadolinium enhancement served as references. Results In healthy volunteers, Multitasking produced motion-resolved fMBV maps with strong MOLLI agreement (bias −0.11%, p=0.84). Subspace harmonization reduced cross-dose variability and aligned cardiac phases. Healthy subjects showed significant diastolic-to-systolic variation (10.55±2.12% vs. 7.73±2.76%, p=0.0004). In IHD patients, ischemic segments exhibited elevated diastolic fMBV with significant systolic compression (16.36±3.44% vs. 10.01±2.24%, p=0.013), remote myocardium showed similar phasic variation (11.16±0.68% vs. 7.98±0.57%, p<0.0001). At the same time, scar demonstrated reduced fMBV with attenuated modulation (7.74±1.95% vs. 5.59±1.97%, p=0.099). Abnormal dynamic fMBV regions corresponded to PET defects and impaired circumferential strain. Conclusion Ferumoxytol-enhanced Multitasking with subspace harmonization enables free-breathing, cardiac-resolved dynamic fMBV mapping, overcoming practical limitations of MOLLI-derived fMBV. This approach enables non-invasive assessment of dynamic myocardial blood volume across the cardiac cycle, with potential utility for characterizing microvascular function in ischemic heart disease.