A new method for objective evaluation of cardiac parametric images.

A new method for analyzing cardiac parametric images is introduced. Its original feature is the representation of local heart motion in a polar coordinate plane which makes the integration of the information contained in three parametric images possible. These images are calculated using first harmonic Fourier filtering and are the maximum volume image, the local ejection fraction image, and the phase image. The last two parametric images are considered as the complex ejection fraction parametric image which is described by a 2D distribution in the complex plane. This method is a generalized 2D version of the well-known phase histogram method. By segmenting the complex plane in which this representation is made, four indexes are defined called the index of hypokinesia, the index of asynergy, the mean phase shift in asynergic regions, and the mean ejection fraction in asynergic regions. The values of those indexes are used to assess the degree and extent of regional wall motion abnormalities. An automated classification as normal, hypokinetic, akinetic, slightly dyskinetic, and strongly dyskinetic made on the basis of the values of those indexes has been compared with a visual classification made by three observers in 219 studies. ROC curves show a good agreement between the automated and visual methods. This new method provides an efficient means of automatically classifying cardiac studies.