Parametric EIT for monitoring cardiac stroke volume

Abstract

The bio-impedance technique appears appropriate for non-invasive cardiac stroke volume (SV) measurement, as the thoracic conductivity distribution is altered during the cardiac cycle due to the heart contraction and blood perfusion. In the present work, the feasibility of a parametric electrical impedance tomography (EIT) for assessing the cardiac SV was studied. An impedance model of the thorax was constructed from segmented axial MRI images along 19 phases of the cardiac cycle. The heart was simulated as an ellipsoid, with its axes' lengths set as the reconstruction parameters, while all other tissues' geometry and conductivity values were kept fixed. A Newton–Raphson parametric optimization scheme was utilized, yielding a correlation between the reconstructed and anatomical left ventricular volumes of 0.97 (p = 2 × 10−11). An analysis of noise sensitivity showed that the proposed algorithm requires an SNR greater than 65 dB. The simulation results were compared to physical data, collected with a portable EIT system (PulmoTrace™, CardioInspect). The validation study was employed for a group of N = 28 healthy patients, and a comparison with impedance cardiography measurements (BioZ®, Cardiodynamics) was made, showing a correlation of r = 0.86 (p = 4 ×10−9). The preliminary results demonstrate that parametric EIT has the potential to measure SV, and may be applicable for both clinical and home environment usage.