comparable imaging of ecg and spect results
and their superposition

A. Mateasik¹, L. Bacharova¹, D. Chorvat Jr.¹, P. Povinec^2
1International Laser Centre,
Ilkovicova 3, 812 19 Bratislava, Slovak Republic
²Teaching Hospital, Comenius University,
Mickiewiczova 13, 813 69 Bratislava, Slovak Republic

Abstract: In our previous work we reported a method allowing a superposition of information obtained by orthogonal ECG in the form of decartograms and by SPECT images, using comparable image techniques. The objective of this study was to adapt this method for optimal superposition of decartograms and SPECT image surfaces with different degrees of regional perfusion defects. The developed software method allows interactive visualization and comparison of decartograms and SPECT-images in 2D and 3D space. A special attention was given to the superposition of decartograms and SPECT images in cases of pronounced perfusion defects on functional 3D surfaces of the heart as defined by SPECT. Deviations from the normal sequence of activated areas in decartograms can be compared with the pathological changes, which are observable in SPECT images. Agreements and disagreements in findings of both methods can be evaluated using fuzzy mathematics.

INTRODUCTION

In our previous work [1-3] we reported a method allowing a superposition of information obtained by orthogonal ECG in the form of decartograms and by SPECT imagings, using comparable image techniques. However, the image surface created on the basis of SPECT data represented the  „normal“ tissue in terms of perfusion as defined by the SPECT method. In cases of pronounced pathology the rendered surface can be different from the real anatomical shape. The objective of this study was to adapt this method for optimal superposition of decartograms and SPECT image surfaces with different degrees of regional perfusion defects.

METHODS

In dipolar electrocardiotopography (DECARTO), orthogonal ECG is transformed by means of a mathematical model to represent the equivalent generator of the cardiac electric field as a uniform double layer with time varying size and location on a spherical surface approximating the ventricular wall [4, 5]. Ventricular activation is represented by time series of maps of activated points, or in the form of summary maps of cardiac excitation on the spherical image surface, so-called decartograms. Decartograms can be visualizing by the projection of the spherical image surface on the various 3D surfaces derived from analytical  geometrical shapes or on the surfaces interpolating a real geometrical properties of the heart (Fig 1). Real surfaces of the heart can be reconstructed from data recorded by standard imaging methods such as SPECT or MRI. In the study we used the geometrical representation of the heart surface computed from a set of cross-sectional images of the heart, which were obtained by single photon emission tomography (SPECT) using myocardial perfusion with talium-201. The SPECT measurements were performed with Siemens imaging system MULTISPECT 2.  The superposition of decartograms and the heart left ventricular surface was done by an approximate merging of separate coordinate system of each surface and follow the projection of DECARTO image sphere onto ventricular surface. The center of coordinate system for SPECT data was assigned to their centroid.

The reconstruction algorithm for creation of 3D surface of the heart was based on the marching cubes approach. If a extended defect in the heart activity is present, the large area of the heart muscle may exhibit no or small metabolic activity. In this case only a surface of metabolic active areas is provided by the reconstruction methods and so regions with undefined topology – the holes may appear on the computed 3D heart surface. Since the principle of the fusion of decartograms and SPECT 3D images of the heart requires the defined surfaces in 3D, we developed methods for recomputation of lacking regions on the SPECT images of injured heart. The holes in the heart functional 3D SPECT surfaces were filled with analytical surfaces (patchs) that were chosen, following the characteristics of set of normal heart SPECT images. These characteristics were assigned to the set of SPECT images after segmentation and clustering analysis of 3D normal heart images.

The comparison of agreements and disagreements in findings of both methods – DECARTO and SPECT can be evaluated using fuzzy mathematics directly on fused images [6,7]. The main input parameters of analysis were the probability of deviation from reference data inherited to both methods and the relative position of activated areas as defined by DECARTO on the SPECT surface of the left ventricle with the respect on the real and recomputerd part of the 3D surface.

Figure 1. The projection of a decartogram onto the spherical surface(I.) and on the 3D surface of the normal heart computed from a set of cross-sectional SPET images.

The Iris Explorer (Numerical Algorithm GDroup, UK) with Workshop C++ 5.0 compilator (Sun Microsystem, USA) compilator were used for development and testing of algorithms for data and images processing.

RESULTS AND DISCUSSION

The developed software method allows a comparison and a fusion of two different informations about the heart activity/status – ECG and SPECT. The superposition of decartograms and cardiac SPECT data is done by the projection of DECARTO data on the approximate surface of left ventricle computed from a set of cross - sectional SPECT images. The computation of left ventricular surface may be interactively control by setting a threshold value which servers as an input parameter for the extraction of the isosurface from volume SPECT data. In the case of incorrect 3D surfaces consequential to the abnormal findings in SPECT data, it is possible to complete the lacking regions or holes by set of help surfaces or patches which are derived from set of 3D surface of normal reference left ventricle [Fig 2]. An evaluation of SPECT and DECARTO data is realized by a comparison with the normal reference values of healthy population and relative position of DEACRTO activated areas on the 3D geometrical surface using fuzzy logic.

The presented combination and analysis of electrocardiographic and SPECT signals using comparable image techniques has a potential to be an investigative and instructional modality for comparative studies of functional and structural changes of the heart. The combined information of these two clinically available methods allows studying the relationship between electrogenesis and the myocardial blood flow and metabolism.

Figure 2:  The projection of a decartogram onto the abnormal 3D surface of left ventricle (I. and II.) with two types of holes (indicated by arrows) and the application of the patch on the surface I (III.).

REFERENCES

[1] L. Bacharova, D. Chorvat Jr, A. Mateasik, M. Cagalinec: „Time dimension in 3D visualisation“, in: Proceedings of the VI International Conference Symbiosis, Szczyrk, Poland, 2001, pp. 1-5.

[2] L. Bacharova, D. Chorvat Jr., A. Mateasik, L.I. Titomir: Topographic presentation of the dipolar content of cardiac electric field. Medizinische Technik, vol. 44, pp. 156–158, 1999.

 [3] A. Mateasik, L. Bacharova, J. Kniz, L.I. Titomir: DECARTO - a tool for superposition of functional and structural characteristics of the heart. Biomedizinische Technik, vol. 46, pp. 79-81, 2001.

 [4] L.I. Titomir, I. Ruttkay-Nedecky: Chronotopocardiography: A new method for presentation of orthogonal electrocardiograms and vectorcardiograms. Int J Bio-Medical Computing, vol. 20, pp. 275 – 282, 1987.

[5] L.I. Titomir, I. Ruttkay-Nedecký, L. Bachárová: Komplexnyj analiz elektrokardiogrammy v ortogonalnych

otvedeniach (Complex analysis of orthogonal electrocardiogram). Moskva, Nauka 2001(in Russian).

[6] I. Rutkay-Nedecky: “Decision making in boundary problems of computerized electrocardiology using fuzzy sets”, in: Computers in Cardiology, IEEE Computer Society Press, Los Alamitos, California, 1992, pp. 557- 560.

[7] L. Bacharova, J. Melotova, I. Ruttkay-Nedecky: Referenčné hodnoty dipólového elektrokardiotopogramu komplexu QRS. (Reference values of dipolar electrocardiotopogram). Bratisl. Lek.Listy, vol. 92, pp. 402–409, 1991(in Slovak, summary in English).