Dynamic Electrical Impedance Tomography Image Reconstruction of Neonate Lung Function based on Linear Kalman Filter techniques Hussein El Dib, Andrew Tizzard and Richard Bayford Department of Natural Sciences, Middlesex University, London, UK. Abstract: Electrical impedance Tomography (EIT) has great potential as a low cost continuous monitoring system for neonate lung function. However EIT data obtained from clinical measurements are inherently noisy and require reconstruction algorithms that are robust enough to cope with these difficulties. In this paper, we examine the potential of a linear Kalman approach for EIT difference imaging. Simulated data is used for the reconstruction process; real time-data is being tested and the results will be introduced at a latter stage of this research. Our objective is to use this initial research to develop a nonlinear Kalman filter for frequency difference imaging of neonate lung function. I INRODUCTION Electrical Impedance Tomography (EIT) has great potential to become an alternative imaging technique to x-ray and MRI. It has advantages compared to other techniques in terms of simplicity, cost, and the absence of ionizing radiation. In the clinical area, there is considerable interest in the use of EIT in imaging lung impedance, in particular, the assessment of lung functionality for neonates, which can only be viably assessed through EIT as there is difficulty in applying traditional imaging techniques on ventilated and sedated babies. Lung growth disorders and maturation are considered as among the most important problems faced by the neonatologist. Premature birth occurs in 5- 10% of all pregnancies and is frequently accompanied by complications due to lung immaturity. Many preterm infants exhibit lung dysfunction characterised by arrested lung development and interrupted alveolarisation. This immature lung phenotype accounts for 75% of early mortality and long-term disability in infants delivered prematurely. Despite improved survival of extremely premature (EP) infants i.e., those born < 27w gestational age (GA), the prevalence of chronic lung disease in infancy (CLDI; commonly defined by oxygen (O 2 ) dependence at 36 weeks post-menstrual age [PMA] i.e., 4weeks before the baby was due to be born), has remained high over the last decade. CLDI is associated with long-term, and possibly life-long, respiratory morbidity. Objective, non-invasive measures of lung maturity and development, oxygen requirements and lung function, suitable for use in small, unsedated infants, are urgently needed to define. At present we are developing an integrated system to image neonate lung function, we have previously examined the effects of boundary form on the reconstruction algorithm for linear difference imaging [1] using a Truncated singular value decomposing (TSVD) which shows that this simple linear approach is sensitive to the change in boundary form. Clinical neonate data is inherently noisy and the application of EIT to the monitoring of neonate lung function would benefit from a reconstruction approach, which is more robust in the presence of noise. A number of research groups have previously used Kalman filtering, [2 - 6], however the filter was applied on data collected from phantom tanks, adults or industrial process, at present the Kalman approach does not appear to be applied to neonates within a real clinical environment. Kalman filtering is an important part of statistical signal processing, namely the statistical estimation theory. Theoretically it is an estimator for what is called linear- quadratic problem; a problem of estimating the instantaneous state of a linear dynamic system perturbed by white noise. The Kalman filter provides a means for inferring the missing information from indirect and noisy measurements; an important behaviour that can that can be employed in EIT dynamic or differential image reconstruction of neonate lung function. It is well known that to solve the image reconstruction problem is to solve an inverse problem and this requires the solution of the forward problem. The evaluation of a Kalman filter based EIT method for image reconstruction of simulated data is presented in this paper. II THEORY The development of the Kalman filter for image reconstruction requires the state space formulation of the inverse problem. In state-space discrete time approach, the temporal evolution of conductivity distribution can be modelled as [2, 9 - 10]: 1 k k k k w σ σ + = + F (2.1) In Eq. (2.1), 1 k σ + is the conductivity distribution at time 1 k + ; N N k × ∈ℜ F , is the state transition matrix at time k and N is number of states of the conductivity distribution; k w is the white Gaussian noise with a known covariance w T N N k k k E ww × ⎡ ⎤ = ∈ℜ ⎣ ⎦ Γ . Because there is no a O. Dössel and (Eds.): WC 2009, IFMBE Proceedings 25/IV, pp. 1643–1645, 2009. www.springerlink.com W.C. Schlegel