Gamma-variate modeling of indicator dilution curves in electrical impedance tomography.

Abstract

Electrical impedance tomography (EIT) is a non-invasive imaging technique, that can be used to monitor regional lung ventilation (V̇) in intensive care units (ICU) at bedside. This work introduces a method to extract regional lung perfusion (Q̇) from EIT image streams in order to quantify regional gas exchange in the lungs. EIT data from a single porcine animal trial, recorded during injection of a contrast agent (NaCl 10%) into a central venous catheter (CVC), are used for evaluation. Using semi-negative matrix factorization (Semi-NMF) a set of source signals is extracted from the data. A subsequent non-linear fit of a gamma-variate model to the source signals results in model signals, describing contrast agent flow through the cardio-pulmonary system. A linear fit of the model signals to the EIT image stream then yields functional images ofQ̇. Additionally, a pulmonary transit function (PTF) and parameters, such as mean transit time (MTT), time to peak (TTP) and area under curve (AUC) are derived. In result, EIT was used to track changes of regional lung ventilation to perfusion ratio (V̇/Q̇) during changes of positive end-expiratory pressure (PEEP). Furthermore, correlations of MTT and AUC with cardiac output (CO) indicate that CO measurement by EIT might be possible.