Article Figures & Data
Figures
- FIG 1.
A, L2 body fracture. Conventional acquired STIR (left) and DL synthesized STIR (right). Note the Syn-STIR manifests fewer motion artifacts, improved sharpness, and greater apparent SNR. B, Sacral metastasis. Conventional Acq-STIR (left) and DL Syn-STIR (right). Note the striking reduction in motion artifacts as well as superior overall quality of the Syn-STIR.
- FIG 2.
Interchangeability when accounting for readers and cases as random effects under the whole-cohort estimated >1000 bootstrapped samples.
- FIG 3.
Interchangeability when accounting for readers and cases as random effects under the trauma cohort estimated >1000 bootstrapped samples. The x-axis represents the estimated interchangeabilities in the unit of percentage, and the y-axis represents the count of bootstrapped samples. A, Trauma class I. B, Trauma class II. C, Trauma class III. D, Mixed effect for all 3 classes.
- FIG 4.
Bland-Altman plots for each tissue from 80 clinical cases. The x-axis represents the mean of the normalized intensity value from Syn-STIR and its nearest neighbor from Acq-STIR, and the y-axis represents the difference between the normalized intensity value from Syn-STIR and its nearest neighbor from the Acq-STIR.
- FIG 5.
Passing-Bablok regression applied to estimate the regression line and intercept. The slope of the disc, CSF, and spinal cord is 1.06, 1.05, and 1.07, which indicates a high correlation between the 2 results. The slope of bone and fat is 0.85 and 0.78, respectively. The results indicate excellent voxelwise consistency between the Acq-STIR and Syn-STIR images.
