An Analysis of Serial Evoked Potentials Using Various Clustering Measures
Mary C. Vespalec · e-Publications@Marquette (Marquette University) · 1989
With the growing use of evoked potentials as an additional clinical tool in diagnosing neurological disorders comes the need for an automated method to classify these waveforms. This thesis improves on previous classification methods and outlines a technique which clusters waveforms using a distance matrix obtained from sampled points of the waveforms. The first chapter gives an overview on the clinical use of evoked potentials, past methods used to classify waveforms and the problem statement. Chapter two describes the signal processing techniques used on the data and lists the steps used to compute the distance matrix. Chapter three discusses the statistical analysis performed on the distance matrix. In chapter four, results from the clustering procedure are shown for both test data and patient data. Lastly, chapter five summarizes the procedure and indicates future areas of analysis.