Table 3 Mean concordance correlation coefficient (CCCs) ± standard deviation between input CS AFt 5.8 and original images, unsupervised output and original images, and self-supervised output and original images. Paired t test results were indicated with **P < 0.01.

From: Deep-learning-based image quality enhancement of compressed sensing magnetic resonance imaging of vessel wall: comparison of self-supervised and unsupervised approaches

 

Input CS AFt 5.8 vs original images

Supervised output

vs original images

Unsupervised output vs original images

Texture features

0.50 ± 0.28

0.47 ± 0.30

0.71 ± 0.20**

First-order features

0.49 ± 0.28

0.43 ± 0.30

0.53 ± 0.35

Wavelet features

0.47 ± 0.28

0.47 ± 0.33

0.62 ± 0.28**

Back to article page