The stroke volume (SV) is an important indicator of the cardiac function. SV can be measured during a Magnetic Resonance Imaging (MRI). The development of a real-time SV estimation technique would benefit the MRI community. We analysed the possibility of an un-calibrated real-time estimate of SV, by using the MRI induced distortion of the Electrocardiogram (ECG) or Magnetohydrodynamic (MHD) effect.
The MHD is induced by the interaction of the blood flow, in the aortic arch, with the MRI static magnetic field. The ECG template was computed during two recordings (inside and outside the MRI), and these two templates were subtracted to estimate the MHD.  ECG data was acquired during a clinical trial, approved by the local ethics comity, and informed consents were obtained from all 9 healthy subjects (4 males, 5 females). ECG were recorded on three leads, in a 1.5T scanner (GE, USA), using a monitoring device (Schiller, France). Phase contrast Cine images were acquired, and processed to extract MRI-based SV measurements.
Gregory, et al. showed the possibility of estimating the MRI-based SV by linearly regressing the MHD from three leads. This process was used to find the four parameters of their linear model. The correlation between MRI-based SV measurements and calibrated MHD SV estimates confirmed their findings.
To assess the feasibility of an un-calibrated estimation, SV was estimated using the mean values across all subjects’ model parameters. The fit was poor (R2=0.3) and with a negative correlation. The population was then divided by gender and by BMI to assess the impact of these factors. However, the fit remained still poor in both cases.
The analysis of the calibration parameters showed a huge variation across the dataset. Gender and body shape could not by themselves explain theses variations. Other factors need to be accounted for un-calibrated SV estimation.