Skip to main content

Table 1 Results on noise robustness of the methods in the two-dimensional model

From: A new approach to the intracardiac inverse problem using Laplacian distance kernel

  HNR MAE (mV) SNR MAE (mV)
  L-Curve
(dB) 25 30 40 50 \(\infty\) 25 30 40 50 \(\infty\)
FTSVD 1.768 1.862 1.269 0.145 1.051 2.266 1.507 8.044 0.192 1.051
FOT 3.094 2.433 1.791 1.599 1.517 4.230 2.734 1.703 1.530 1.518
FTV 1.766 1.066 0.366 0.117 8.11\(e^{-12}\) 4.032 2.233 0.697 0.226 8.11\(e^{-12}\)
  LOO
FTSVD 7.846 7.97 8.015 8.026 8.011 8.030 8.017 8.013 8.011 8.011
FOT 1.363 0.867 0.335 0.116 4.58\(e^{-13}\) 1.188 0.900 0.499 0.210 4.58\(e^{-13}\)
FTV 1.388 0.805 0.258 0.083 7.88\(e^{-12}\) 2.112 1.197 0.380 0.117 7.88\(e^{-12}\)
  Grid search
DSM−SVR 1.754 1.053 0.354 0.107 3.28\(e^{-4}\) 4.030 2.232 0.696 0.225 3.28\(e^{-4}\)
  1. MAE has been calculated for all the tested methods as a function of SNR and of HNR, and also when using L-Curve versus the LOO with traditional algorithm for the free-parameter search and Grid search for DSM−SVR