Table 2 Performance comparison of different CNN architectures based on validation set A z and test set Dice score considering observer 1 and observer 2 as the reference standard.

From: Location Sensitive Deep Convolutional Neural Networks for Segmentation of White Matter Hyperintensities

Method

Without location features

With location features

Validation set A z

Test set Dice (obs1)

Test set Dice (obs2)

Validation set A z

Test set Dice (obs1)

Test set Dice (obs2)

SS

0.9939

0.731

0.729

0.9972

0.781

0.778

MSEF

0.9947

0.762

0.752

0.9966

0.777

0.769

MSIW

0.9966

0.778

0.768

0.9972

0.795

0.787

MSWS

0.9965

0.773

0.760

0.9973

0.792

0.783

Back to article page