The establishment of full body splicing radiograph and study on its accuracy and reliability of global sagittal balance

English Keywords:Full-body splicing radiograph  Global sagittal balance  Pelvic sagittal parameters  Reproducibility of results

English Abstract:

  【Abstract】 Objectives: To construct a new full body splicing radiograph through digital radiography(DR) system, and to discuss the reliability and accuracy in its evaluation of overall sagittal balance. Methods: From October 2018 to October 2019, the DR system was applied to collect the image data of 30 volunteers at one time continuous exposure, and the medical image processing software was used to automatically stitch together and obtain clear and accurate full body splicing radiograph. There were 13 men and 17 women with an age of 45-65 years(mean age 56.5±6.1 years). The volunteers were also subjected to an examination of whole spine lateral radiographic and lateral scanogram of the normal DR radiography system. The parameters of thoracic kyphosis(TK), lumbar lordosis(LL), pelvic incidence(PI), pelvic tilt(PT), sacral slop(SS), sagittal vertical axis(SVA), sacrofemoral angle(KFA), knee flexion angle(KA), ankle angle(AA), and pelvic shift(P.shift) were measured. Two spine surgeons and an imaging physician performed imaging measurements by using Surgimap software at three different occasions. The correlation coefficients were validated by independent sample t-test. The consistency of parameter values within and between the observers was analyzed by the intragroup correlation coefficient(ICC). Results: The independent sample t-test of the full-body sagittal splicing radiograph and the whole spine radiograph and lower limbs extremity had no statistical difference(independent sample t-test, 0.782-0.969 P>0.05). The full-body sagittal splicing radiograph was compared to the whole-spine lateral radiograph, and the intraobserver ICCs of TK, LL, PI, PT, SS, SVA were 0.696-0.857, 0.724-0.873, 0.672-0.943, 0.691-0.851, 0.648-0.852, and 0.645-0.867, respectively. The interobserver correlation coefficients were 0.678-0.822, 0.704-0.872, 0.772-0.904, 0.763-0.936, 0.665-0.824, and 0.541-0.665. The full-body sagittal splicing radiograph was compared to the lateral scanogram, and the intraobserver ICCs of PI, PT, SS, KFA, KA, AA, and P.shift were 0.673-0.932, 0.623-0.828, 0.634-0.861, 0.701-0.873, 0.645-0.867, 0.679-0.855, and 0.592-0.827, respectively. The interobserver correlation coefficients were 0.665-0.914, 0.631-0.811, 0.625-0.843, 0.541-0.765, 0.591-0.753, 0.613-0.798, and 0.543-0.762. Intraobserver and interobserver reproducibility for both modalities were good to excellent. Conclusions: It is feasible to construct the full body splicing radiograph by digital radiography system. The sagittal parameters of the spine-pelvis-lower extremity measured by it are similar to the measurement results on the whole-spine lateral radiograph and the lateral scanogram, and have good reliability and repeatability.