Show simple item record

dc.contributor.authorKozey, Cheryl Lynne.en_US
dc.date.accessioned2014-10-21T12:38:17Z
dc.date.available1992
dc.date.issued1992en_US
dc.identifier.otherAAINN76689en_US
dc.identifier.urihttp://hdl.handle.net/10222/55298
dc.descriptionThis study investigated the value of electrocardiographic (ECG) body-surface mapping during sinus rhythm for noninvasively identifying individuals at risk of life-threatening ventricular arrhythmias. Training and test sets were formed from 255 subjects--51 normals, 102 patients with recurrent ventricular tachycardia (VT) and 102 patients with myocardial infarction (MI) but no clinical arrhythmias. Measurements analyzed for each subject consisted of 117 QRST-integral values calculated from as many ECG leads; the rationale for choosing QRST-integral maps was that they reflect the distribution of ventricular repolarization properties, any disparity of which might constitute an arrhythmogenic substrate. The Karhunen-Loeve (KL) technique was applied to represent each subject's measurements by 16 coefficients, and the Kittler and Young (KnY) method of feature selection further reduced the number of characteristic features to only one or two. Nondipolar content of QRST-integral maps, determined from higher-order KL coefficients, was significantly lower for the normal subjects than for the patient groups, but it differentiated poorly between the latter groups. A much-improved diagnostic classification was based upon discriminant analysis which used three selected KL features or, alternatively, just one KnY feature. The best classification results were achieved with one KnY feature derived from the two patient groups of the training set; the diagnostic performance (percentage of patients classified correctly) was 89% for the training set and 85% for the test set. The errors to be expected in classifying future observations were estimated by bootstrap and cross-validation methods. Furthermore, the potential clinical usefulness of all classification procedures as methods for noninvasively evaluating the effectiveness of drug therapy was assessed. The results of this study demonstrate that multiple-lead body-surface ECGs contain valuable information that, if properly extracted, can identify an arrhythmogenic substrate in the myocardium of patients at risk of malignant ventricular arrhythmias.en_US
dc.descriptionThesis (Ph.D.)--Dalhousie University (Canada), 1992.en_US
dc.languageengen_US
dc.publisherDalhousie Universityen_US
dc.publisheren_US
dc.subjectEngineering, Biomedical.en_US
dc.subjectHealth Sciences, Medicine and Surgery.en_US
dc.subjectHealth Sciences, Pathology.en_US
dc.titleBody-surface potential mapping: Classification of spatial features associated with an arrhythmogenic state.en_US
dc.typetexten_US
dc.contributor.degreePh.D.en_US
 Find Full text

Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record