Optimized Positioning of ECG Electrodes for WSN Applications - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 197

For every combination
of 3 differential leads

One combination
of 3 differential
leads

Differential
Leads

Target 12-
lead ECG

MECG Solve MLR model

Reconstructed
12-lead ECG

Calculate the correlation
coefficient between
reconstructed and target
12-leads ECGs

Correlation
coefficients

Find combination of 3 differential
leads that result in the best
correlation

Optimal
diferential leads

Fig. 9.11 Optimal leads selection with Multivariate Linear Regression (MLR)

In the case of PCA, the transformation matrix is the matrix defined by PCs, and in
the case of MLR the transformation matrix is composed of the coefficient vectors.
By studying a sufficiently large number of cases for different patients it
should be possible to find optimal leads that will be suitable for most patients
and to calculate a global transformation that gives, on average, the best possible
fit for each individual case [ 31 ]. Although possible, such an approach is not
necessary in WSN applications because wireless electrodes have a lot of
positional freedom, since they are not limited by the connection wires of an
ECG device.