Table 7 Brain structure segmentation results of TMS on Iseg2017. “T2 + \(\widehat{T}1\)” in bold font indicates our method (TMS) where inputs are both T2 and \(\widehat{T}1\) images. “T2” indicates the traditional FCN method where inputs are only T2 images. “T2 + T1” indicates the upper bound.

From: MRI Cross-Modality Image-to-Image Translation

 

Dice(wm)

Δ

Dice(gm)

Δ

T2

0.649

0.767

T2 + \(\widehat{{\bf{T}}1}\)

0.669

3.08%

0.783

2.09%

T2 + T1

0.691

0.797