Automatic analysis of gated heart studies using a Galois lattice

Abstract

Gated heart studies are currently assessed using the parametric images of first harmonic phase and amplitude in addition to the original images. More information can be obtained by multi-harmonic analysis, but the quantity of information is too great to be assessed by visual methods alone. We propose an automatic diagnosis support system which will use all the additional information, compress and analyse the data, and present the results in a single image to assist the consultant in his or her diagnosis. Multi-harmonic analysis produces many parameters to describe features of the images. A set of parameters will be selected using guidelines that link parameters and intervals of their values to clinical conditions. Images are processed using these parameters and the results stored in a binary matrix. A Galois lattice structure is then derived from the matrix where a vertex of the lattice represents a ‘significant’ region of the heart together with a ‘logic description’ of this region. Using this, areas of clinical interest can be found and displayed on a diastolic image of the heart.

Key Words:

• Expert system
• Cardiac
• Image segmentation
• Lattice