Thoracic EIT in 3D: experiences and recommendations

No Thumbnail Available
Grychtol, Bartłomiej
Schramel, Johannes Peter
Braun, Fabian
Riedel, Thomas
Auer, Ulrike
Mosing, Martina
Braun, Christina
Waldmann, Andreas D
Böhm, Stephan H
Adler, Andy
Objective: In EIT applications to the thorax, a single electrode plane has typically been used to reconstruct a transverse 2D ‘slice’. However, such images can be misleading as EIT is sensitive to contrasts above and below the electrode plane, and ventilation and aeration inhomogeneities can be distributed in complex ways. Using two (or more) electrode planes, 3D EIT images may be reconstructed, but 3D reconstructions are currently little used in thoracic EIT. In this paper, we investigate an incremental pathway towards 3D EIT reconstructions, using two electrode planes to calculate improved transverse slices as an intermediate step. We recommend a specific placement of electrode planes, and further demonstrate the feasibility of multi-slice reconstruction in two species. Approach: Simulations of the forward and reconstructed sensitivities were analysed for two electrode planes using a ‘square’ pattern of electrode placement as a function of two variables: the stimulation and measurement ‘skip’, and the electrode plane separation. Next, single- versus two-plane measurements were compared in a horse and in human volunteers. We further show the feasibility of 3D reconstructions by reconstructing multiple transverse and, unusually, frontal slices during ventilation. Main results: Using two electrode planes leads to a reduced position error and improvement in off-plane contrast rejection. 2D reconstructions from two-plane measurements showed better separation of lungs, as compared to the single plane measurements which tend to push contrasts in the center of the image. 3D reconstructions of the same data show anatomically plausible images, inside as well as outside the volume between the two electrode planes. Significance: Based on the results, we recommend EIT electrode planes separated by less than half of the minimum thoracic dimension with a ‘skip 4’ pattern and ‘square’ placement to produce images with good slice selectivity.
Publication Reference
Physiological Measurement, vol. 40 (7), pp. 074006