Table 3 Root mean squared error (RMSE) per lead comparing real and ECGs reconstructed with GAN

From: Evaluating the feasibility of 12-lead electrocardiogram reconstruction from limited leads using deep learning

 

Lead I

Lead II

V1

V2

V3

V4

V5

V6

1-lead reconstruction (μV)

–

79

76

136

136

120

96

79

2-lead reconstruction (μV)

–

–

72

134

135

110

82

64

SD real

122

136

145

239

231

235

212

166

  1. Average root mean squared error (RMSE) for every lead of the reconstructed ECGs. The results were derived from the GAN models trained using both 1-lead and 2-lead configurations.
  2. SD standard deviation.
Back to article page