Table 5.
SP | LLC | ULC | ||||
---|---|---|---|---|---|---|
Mean ± SD | %CV | Mean ± SD | %CV | Mean ± SD | %CV | |
3D FSPGR | 1.63 ± 0.29 | 17.70% | 0.77 ± 0.13 | 17.10% | 0.81 ± 0.11 | 13.50% |
T1WIDL | 1.67 ± 0.15 | 8.70% | 0.73 ± 0.07 | 9.50% | 0.76 ± 0.07 | 9.70% |
vs. T1WIO (p value) | 0.007 | 0.653 | <0.001 | |||
vs. 3D FSPGR (p value) | 0.428 | 0.109 | 0.02 | |||
T2WIDL | 1.81 ± 0.10 | 5.30% | 0.95 ± 0.07 | 7.10% | 0.93 ± 0.08 | 8.40% |
vs. T1WIO (p value) | <0.001 | 0.917 | 0.979 | |||
vs. 3D FSPGR (p value) | <0.001 | <0.001 | <0.001 | |||
T1WIO | 1.56 ± 0.18 | 11.50% | 0.72 ± 0.11 | 15.40% | 0.67 ± 0.09 | 13.80% |
vs. 3D FSPGR (p value) | 0.201 | 0.094 | <0.001 | |||
T2WIO | 1.70 ± 0.16 | 9.10% | 0.95 ± 0.13 | 13.50% | 0.93 ± 0.10 | 10.80% |
vs. 3D FSPGR (p value) | 0.169 | <0.001 | <0.001 |
Note. The result of the thickness measured on the axial T1WI, T2WI, and 3D FSPGR images were compared between deep-learning-based reconstruction images and conventional reconstruction images, deep-learning-based reconstruction images and 3D FSPGR images, and conventional reconstruction images and 3D FSPGR images. T1WIO = original T1-weighted FSE images, T1WIDL= deep learning–reconstructed T1-weighted FSE images, T2WIO = original T1-weighted FSE images, T2WIDL = deep learning–reconstructed T2-weighted FSE images, 3D FSPGR= three-dimensional fast spoiled gradient-recalled images. SD = standard deviation; %CV = percent coefficient of variation (SD/mean). SP = septal cartilage, ULC = upper lateral cartilage and LLC = lower lateral cartilage. Significant p values are expressed in bold.