Sliding Window Based Machine Learning System for the Left Ventricle Localization in MR Cardiac Images

Table 1

Final training parameters for BPNN.

Network parameters	Values
Number of training samples; Number of input neurons	250; 1600
Number of hidden neurons	80
Activation function at hidden and output layers	Log-Sigmoid
Learning rate ()	0.11
Momentum rate ()	0.80
Required error (MSE)	0.01
Epochs	1,215
Training time	40 secs