Table 1 Application of the proposed method to correct SA for 70 kidneys: 46 with no or mild SA, 14 with moderate SA, and 10 with severe SA.

From: Accurate exclusion of kidney regions affected by susceptibility artifact in blood oxygenation level-dependent (BOLD) images using deep-learning-based segmentation

 

Kidneys with mild SA

Kidneys with moderate SA

Kidney with severe SA

All the kidneys

RMS (voxels)

18.1 ± 6.4

27.2 ± 8.6

51.5 ± 13.1

29.2 ± 16.4

Relative RMS (%)

0.19% ± 0.07%

0.31% ± 0.08%

0.58% ± 0.17%

0.32% ± 0.19%

SA-excluded ROI DICE

94.3% ± 3.8%

92.8% ± 4.7%

94.5% ± 1.7%

93.9% ± 3.4%

R2* decrease after SA correction (s−1)

1.6 ± 0.9

4.7 ± 1.8

8.5 ± 2.8

4.3 ± 3.4

R2* decrease after SA correction (%)

5.1% ± 2.7%

14.7% ± 5.3%

23.4% ± 7.6%

12.6% ± 9.1%

  1. The performance of the method was evaluated based on the goodness of fit (RMS and relative RMS) of bilinear fitting, DICE of ROI as compared to that by the manual method, and the averaged R2* values over kidney ROI.
Back to article page