Interactive biosignal analysis and classification

We often encounter experimental results that are difficult to quantify although the visual recognition of a phenomenon is apparent. We are able to detect patterns and features with our vision that are complex to specify algorithmically. This is because our brain has evolved into a powerful pattern matching machine. One important class of measurement results in medical research are signals that describe a cycling pattern of some phenomenon. In this work, we developed a software tool, Anomaly Explorer, to assist in classification of Ca2+ signals emitted by cardiomyocytes. Anomaly Explorer was designed and implemented to analyze and classify Ca2+ signals with criteria similar to the criteria used by laboratory personnel. The software tool is based on interactive visualization with a direct manipulation user interface. Our Anomaly Explorer has proven to be a valuable tool as it makes the analysis faster, more accurate, and person-independent. The definition of signal anomalies in terms of visual properties is more flexible than numerical or analytical specification. Our approach is not limited to the Ca2+ signal analysis. The anomaly types are domain-specific, but the interaction and visualization ideas can be applied in any signal analysis problem.