BSPM spatial data reduction by normalized Karhunen-Loeve expansion

The authors introduce a normalized Karhunen-Loeve (K-L) expansion in the spatial data reduction of BSPM (body surface potential map) data. In the normalized K-L expansion, the covariance matrix is calculated by the normalized vectors (maps), and the purpose of the normalization is to eliminate the energy inhomogeneity. The results show the superior performance of the normalized K-L expansion over the original K-L expansion in both RMS errors and correlation coefficients.<
>