Cardiac functional analysis system employing gradient image segmentation

Abstract

A non-invasive imaging device obtains a four dimensional (4D) image data set of a living subject representing three dimensions and time in relation to the subject's cardiac cycle. In order to determine various vascular parameters, it is useful to segment the image data set into internal structures defined as having the same tissue types contiguous locations. To accomplish this, a gradient calculation unit constructs a gradient data set from the image data set indicating the magnitude of spatial changes in the image data set. A plurality of locations are selected by an operator with a pointing device during `training` along with corresponding data values in the image and gradient data sets. These data values are plotted by a scatter generator against each other to construct a scatter plot then processed to determine a bivariate statistical probability distribution. The remaining data values are then assigned a tissue type based upon their plot on the bivariate statistical probability distribution. Contiguous locations having the same tissue type assignment are identified as a solid structure by a connectivity calculator. These solid structures may be the internal volume of cardiac chambers. Since these volumes may be accurately measured over the cardiac cycle, a functionality calculator determines vascular functionality, such as ejection fraction, and cardiac output and displays the results on a monitor to the operator.