Isotropic versus anisotropic techniques in cardiac computed tomography images processing

Abstract

The objective of the work is to compare the performance of two filters, one isotropic and another one of anisotropic diffusion based on gradient. To do this, experiments are carried out to establish which of the filters exhibits a better behavior against the imperfections that characterize the computed tomography images. The structure of the experiments is as follows: a) The parameters linked to the aforementioned filters are identified. b) The ranges of valuesof these parameters and the way to use them are established. c) A database of three-dimensional cardiac images is filtered by applying, independently, the aforementioned filters considering a pre-established subset of values associated with the parameters. d) All the filtered images are addressed by a segmentation process, based on the growth of regions, which allows extracting the 3D morphology of the thoracic external aorta. e) As a metric to evaluate the performance of each technique, the Jaccard similarity index (JSI) is used. f) The one that generates the lowest calculated JSI is selected as the best technique when comparing a reference segmentation with all generated segmentations. The results indicate that the anisotropic diffusion filter, based on a gradient, obtained the best performance.