Free-breathing estimation of total circulating blood volume using ferumoxytol-enhanced MRI on a commercially available 0.55T platform

Background: Quantification of total circulating blood volume (TBV) provides clinically valuable information for the management of patients with cardiovascular and systemic diseases, including heart failure, anemia, sepsis, and disorders associated with fluid overload.^1,2 Conventional reference methods, such as carbon monoxide rebreathing, dye dilution, and nuclear tracer techniques, are invasive, imprecise, or limited by assumptions about vascular compartmentalization.^3,4 CMR with intravascular contrast agents such as ferumoxytol has emerged as a promising noninvasive alternative, with prior studies demonstrating feasibility at 1.5T under breath-hold conditions.⁵ Importantly, low-field CMR at 0.55T has shown advantages in contrast sensitivity, reduced susceptibility artifacts, and broader accessibility.^6-8 However, TBV quantification using ferumoxytol at 0.55T and, critically, under free-breathing conditions has not been investigated.

Methods: We performed in vivo experiments on domestic swine (n = 11, weight: 45.0 ± 4.4 kg) using a 0.55T platform (MAGNETOM Free.Max, Siemens Healthineers, Erlangen, Germany) under a protocol approved by local IACUC. After localization of a mid-ventricular short-axis slice and acquisition of pre-contrast T1 maps, animals received intravenous infusion of ferumoxytol (Feraheme, Azurity Pharmaceuticals, Woburn, MA) at a dose of ~1 mg/kg. After the infusion was completed, T1 mapping was performed every 10-20 minutes for up to 60 minutes. We used two T1 mapping techniques: (a) free-breathing, SASHA-like^9,10 approach using a saturation-recovery(SR) bSSFP sequence, (b) standard breath-hold MOLLI sequence.^11,12 T1 values from the blood pool ROIs (in the left ventricular cavity) were extracted at each time point. TBV was estimated using a multi–time-point linear fitting approach similar to prior work^5,13, where the logarithm of the distribution volume is plotted against time, and the y-intercept of the fitted line gives estimated TBV. The TBV estimation method and the CMR protocol are described in Fig. 1.

Results: As shown in the representative case in Fig. 2, SR-bSSFP and MOLLI datasets demonstrated progressive blood pool T1 recovery following ferumoxytol infusion. Mean TBV was 89.9 ± 3.8 mL/kg for SR-bSSFP and 91.0 ± 2.6 mL/kg for MOLLI (p=n.s.), with no statistically significant difference (Fig.3). Both values were within the expected physiological range for swine reported in the previous CMR studies (85–95 mL/kg).⁵ Importantly, free-breathing SR-bSSFP yielded stable TBV quantification with variance comparable to MOLLI, suggesting resilience to respiratory motion.

Conclusion: This study demonstrates the feasibility of FE TBV quantification at 0.55T using free-breathing SR-bSSFP T1 mapping. These findings highlight the potential of low-field CMR combined with free-breathing imaging to provide accurate and clinically relevant TBV measurements, offering a noninvasive and practical tool for clinical applications.

*S Zeynali and HB Unal contributed equally to this work.

Figure 1: Description of the methodology for estimating total blood volume (TBV) using ferumoxytol-enhanced CMR. (A) Similar to other contrast agents, ferumoxytol has a defined half-life and gradually clears from blood circulation. This clearance is observed as a recovery of post-contrast blood T1 toward its native value, which can be monitored with T1 mapping (left plot). Given a known administered dose, changes in post-contrast R1 (1/T1) relative to native R1 enable estimation of TBV with a closed-form formula. However, these estimates tend to exceed true TBV due to the assumption of constant contrast concentration. To correct this, the logarithm of volume estimates is fitted to a line, and the y-intercept yields the true TBV. (B) The in vivo protocol begins with localization of a short-axis slice and acquisition of native T1 maps. Ferumoxytol is then slowly infused over ~15 minutes. Free-breathing SR-bSSFP T1 maps are acquired every 10–20 minutes for about an hour. Breath-hold MOLLI T1 maps (not shown) are also acquired at matching time points for comparison. TBV is estimated from T1 values within ROIs in the left ventricular blood pool using the method described in panel A.
Figure 2: Representative swine study demonstrating total blood volume (TBV) estimation from serial ferumoxytol-enhanced T1 mapping. (A) Free-breathing SR-bSSFP–based T1 maps at baseline (native), and at 16 and 24 minutes post-infusion, with blood pool ROIs indicated with corresponding T1 value are shown. As expected, blood T1 drops dramatically following ferumoxytol infusion. Also, the blood T1 increases from t=16 min to t=24 minutes as a result of ferumoxytol clearance. (B) (Left) Recovery of blood pool T1 values from the free-breathing SR-bSSFP T1 mapping over ~60 minutes following ferumoxytol infusion. (Right) Linear fitting of the logarithm of the normalized distribution volume V′(t) vs. time. The y-intercept yields the estimated TBV. (C) Corresponding MOLLI-derived measurements from the same animal for validation, showing comparable T1 recovery kinetics and TBV estimation.
Figure 3: Comparison of total blood volume (TBV) estimates derived from free-breathing SR-bSSFP and breath-hold MOLLI T1 mapping. Group-averaged TBV was 89.9 ± 3.8 mL/kg for SR-bSSFP (free-breathing) and 91.0 ± 2.6 mL/kg for MOLLI (breath-hold), with no statistically significant difference demonstrating the agreement between the two methods. These results validate the feasibility of free-breathing SR-bSSFP acquisitions for TBV quantification, with accuracy comparable to the standard breath-hold MOLLI approach.