J.R. Boston, G. Akyol, and L. Baloa (USA)
fuzzy logic, signal detection, uncertainty, QRS detection, cost analysis.
A QRS detection system was developed that uses fuzzy logic to combine the outputs of two QRS detector algorithms. The two algorithms are sensitive to different types of noise, producing detection errors in different situations. Combination of the outputs was implemented using several methods of fuzzy aggregation, including the minimum operator, non compensatory operators, and compensatory operators, and evaluated using a modification of Bayes cost analysis. The fuzzy detector incorporates a measure of uncertainty and can conclude that a QRS is present, that a QRS is not present, or that no decision can be made. Concluding that a point is uncertain can avoid an outright error by identifying for further analysis points having higher a priori evidence, compared to an arbitrary time point in the waveform, that a QRS complex exists. Performance varied for different types of aggregation operators, generally showing a trade-off between the number of errors and the number of uncertain classifications. Using the minimum aggregation operator, error rates were reduced substantially below those obtained with the individual algorithms, but at a cost of a large number of uncertain detections. Classification performance was evaluated for different cost functions. Performance was relatively insensitive to the cost function for aggregration operators close to the minimum operator.
Important Links:
Go Back
