摘要 |
A noise model14 is derived from a first portion of a cardiac electrogram 11 where a physiological signal is presumed to be absent. A second portion of the electrogram 12, presumed to contain a physiological signal, is transformed into the noise model 14. The physiological signal is identified as those portions of the electrogram that do not conform to the noise model 14. The noise model 14 is created by correlation of the noise regions 11 with templates 13. The physiological signal regions 12 are correlated with the templates 13 and projected into eigenvector space of the model. Results within the space indicate the signal is noise, but results exceeding the limits of the space indicate the signal is physiological. Risk of sudden death due to cardiac arrhythmias can be predicted by observing paced endocardial electrograms, specifically detecting small deflections in the electrogram following early stimulation. These small individual potentials can be detected despite the presence of noise. |