Use of an anatomical atlas in real-time EIT reconstructions of ventilation and pulsatile perfusion in preterm infants.

Electrical impedance tomography (EIT) is a bedside imaging technique in which voltage data arising from current applied on electrodes is used to compute images of admittivity in real time. Due to the severe ill-posedness of the inverse problem, good spatial resolution poses a challenge in EIT. Conversely, the temporal resolution is high, facilitating dynamic bedside imaging. In this work, we propose a real-time linearized reconstruction algorithm that makes use of an anatomical atlas to provide prior spatial information at two stages of the reconstruction with the goal of improving the spatial resolution. The algorithm updates a non-constant initial estimate of an anatomically relevant distribution of conductivity and susceptivity obtained from the mean of the atlas, and using the Schur complement method as a post-processing technique. Two atlases are constructed from a database of CT scans of 89 infants; one for the reconstruction of ventilation and one for the reconstruction of pulsatile perfusion. The algorithm is applied to data collected on 16 premature infants with lung disease of prematurity and 5 healthy control infants to reconstruct conductivity and susceptivity images of both ventilation and pulsatile perfusion in real time using the ACT 5 EIT system. EIT parameters describing homogeneity of ventilation distribution throughout the lung and the distribution anterior/posterior and in the left versus right lung were computed for each infant. The left/right ventilation distribution was found to distinguish between the healthy and the preterm infants with statistical significance (p-value< 0.05). The reconstructions demonstrate qualitatively improved resolution when compared to the NOSER algorithm currently used on the ACT 5 system for real-time bedside imaging, and the ability to image changes due to ventilation and pulsatile perfusion, as well as regional inhomogeneity. Since CT scans were not available for these infants, there is no gold standard for validation. In conclusion, we present a novel real-time algorithm with the goal of improving spatial resolution for bedside imaging with EIT for conductivity and susceptivity imaging of ventilation and pulsatile perfusion, with the potential to aid in the evaluation of lung function in infants at the bedside.