Table 2 Classification accuracy evaluation metrics with median and interquartile range (IQR) in parenthesis of the CNN classification performance over 100 runs of the data in addition to the results with the clinical interpretation with the hold-out (test) dataset.

From: Classification of ischemia from myocardial polar maps in 15O–H2O cardiac perfusion imaging using a convolutional neural network

Method

ACC

AUC

F1S

SEN

SPE

PRE

CNN

0.8261 (0.0652)

0.8058 (0.0654)

0.7647 (0.1014)

0.6500 (0.1000)

0.9615 (0.0000)

0.9286 (0.0103)

Clinical

0.8696

0.8558

0.8333

0.7500

0.9615

0.9375