Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Second-Order blind identification of underdetermined mixtures
ICA'06 Proceedings of the 6th international conference on Independent Component Analysis and Blind Signal Separation
| Hi-index | 0.00 |
The paper discusses theoretical problems that arise in extracting fetal ECG (fECG) from maternal ECG (mECG) and especially the diagnostically important fetal P-waves and T-waves early in pregnancy for beat-to-beat diagnosis. Such extraction is important for identifying fetal cardiac disorders. This problem illustrates the inadequacy of popular extraction methods, especially ICA (Independent Component Analysis, including underdetermined ICA) and other BSS (Blind Signal Separation) methods to a large class of problems (medicine, speech processing) where either extraction of a single signal or of multitude of signals that are buried in a single observation is concerned or when the number of source signals exceeds the number of observations and especially when extraction of a very weak source signal is required under the above conditions. Furthermore, the paper illustrates the power of the feature-based BAF (Blind Adaptive Filtering), to those problems. Analysis is given as are extraction results for beat-wise extraction fECG and fetal P and T waves, to highlight both the problem and the BAF solution.