CompanyName: University of Minnesota
Abstract: Evaluation of Two EIT Reconstruction
Algorithms Using a 3D Finite Difference Model of the Thorax
Robert P. Patterson and Jie Zhang
University of Minnesota
The EIT reconstruction algorithm developed by the Sheffield group (B. Brown et al) was compared to algorithm developed by the Russian EIT group (A. Korjensvsky et al.) using a 3D finite difference model of the thorax. The model was constructed from 40 transverse MRI slices 0.5 cm thick gated with the ECG. The model had 225,000 control volumes. In the evaluation only end diastole and systolic images were compared. The Sheffield back projection method only produces dynamic images. The Russian algorithm produces static images. Dynamic images were created from the static images by subtracting two static images produced at different times or under different simulated conditions. The reconstructed images using the Russian algorithm were obtain by sending data to their web site (http://www.cplire.ru/tserver/).
The reconstructed images were compared under different conditions: end diastole with end systole; by varying individually the resistivities in lungs and ventricle; and by changing both the size and position of the 16 electrodes used around the thorax.
The images created by changing from end diastole to end systole were similar to those obtained directly from human subjects. The isolated changes in the lungs, ventricles, and electrode size and location produced unexplained significant deviations in the images. For example increases in lung resistivity produced the expected changes in the lung region but also a significant conductivity decrease in the cardiac region.
The images from the Russian EIT group were smoother and had less unexplained small regional changes and noise compared to the Sheffield images.