Application of the statistical method of cluster analysis as a pattern recognition technique wm investigated by classification of cardiowascular tissues on the basis of their trace metal composition. The distribution patterns of 13 trace metals among 17 tissues of the heart and its appended blood vessels were determined. Cluster analysis of the data resulted in ready differentiation of tissues derived from ordinary and specialized myocardium from those of blood vessels and heart valves. Further similarities and differences among the 1'7 tissues were identified by comparing the patterns generated when different numbers of groups (i.e., 3, 4, 5, 6, and 7 groups) were dcsilfrxrted. It is suggested that cluster analysis may provide a generally useful tool in the interpretation of data from multi-phasic laboratory testing ptoccdures.
Research on trace metals has been stimulated in recent years by the increasing awareness of the wide variety of biological effects mediated by trace metals and by the availability of analytical techniques sensitive in the range of p-pm_ and less. In many studies it is desirable to determine several trace metals simultaneously thus producing profiles of trace metal composition. Such studies commonly generate large amounts of numerical data that need to bc manipulated to allow quantitative comparison among
the objects studied. In our studies emission spectrometric procedures have been employed for the simultaneous determination of up to 17 elements in biological specimens 1 l-5]. Recently, to facilitate the handling of extensive sets of analytical data, statistical methods of pattern recognition have been exploited [ 6, 'i] _ Such statistical procedures offer the promise of ready quantitative comparisons of trace metal profiles of different objects, and indeed should