Abstract 15125: Non-Invasive Blood Oxygen Saturation Measurement by a Calibration-Free MRI Method
Non-invasive oxygen saturation (O2 sat) measurement in the cardiac chambers and blood vessels would be clinically valuable in congenital heart disease, heart failure and pulmonary hypertension. A calibration-free magnetic resonance (MR) technique for accurate, patient-specific O2 sat measurement was developed and compared against the invasive reference method.
The dependence of blood O2 sat on the effective transverse relaxation time (T2) is characterized by the Luz-Meiboom (L-M) model. It includes patient-specific parameters such as hematocrit and O2 sat, and MR-specific parameters such as field strength and inter-echo spacing. Additional unknown parameters characterize physical and magnetic blood properties. We hypothesize that O2 sat can be determined accurately from this complex model by acquiring a set of effective blood T2 values at distinct inter-echo spacings, and jointly fitting the T2 data to the L-M model to estimate all unknown parameters, including O2 sat.
A porcine model (n = 7) was studied with graded hypoxemia across a range of O2 sat levels. To generate a set of effective blood T2 measurements, four distinct blood T2 maps were acquired (~3 minutes) at each hypoxemia stage. O2 sat estimated using two forms of the L-M model with four (Model 1) and six (Model 2) unknown parameters was compared to reference O2 sat measured from invasive catheterization samples.
The linear regression and Bland Altman plots in Figure 1 show O2 sat estimated by the proposed technique against the reference measurement. O2 sat estimation with Model 1 demonstrated a higher correlation (1a, R2 = 0.90, p = 3.1e-14) with the reference method compared to Model 2 (1c, R2 = 0.83, p = 3.7e-11). The bias with the reference method was not significant for Model 1 (1b, -0.9 ± 3.7%, p =0.20) and Model 2 (1d, -1.7 ± 6.1%, p = 0.15).
A calibration-free MRI method has been presented for non-invasive measurement of O2 sat. Results from a porcine model show high correlation with invasive blood gas analysis.
Author Disclosures: J. Varghese: None. R. Ahmad: None. L.C. Potter: None. O.P. Simonetti: Research Grant; Modest; Siemens Healthcare.
- © 2016 by American Heart Association, Inc.