Linear prediction of canine gastric signals

Abstract

Recordings of gastric electrical activity from the abdominal surface are generally contaminated by interferences from other biological sources and by additive noise, which makes a routing interpretation difficult. The technique of linear prediction has been used to enhance the quality of the embedded gastric signal component. The results indicate that linear prediction is capable of separating the gastric electrical signal component from uncorrelated interferences. Moreover, this approach has the intrinsic ability to track irregularities of the gastric activity, such as tachygastria, which is demonstrated by examples.<>