3D quasi-automatic spine length assessment using low dose biplanar radiography after surgical correction in thoracic idiopathic scoliosis

Abstract

ObjectiveSurgical correction of thoracic scoliosis leads to a height improvement. Our objectives were to assess how the linear and developed spinal column lengths relate to the frontal and sagittal parameters after a surgical correction of thoracic idiopathic scoliosis, and whether the measurement of these lengths is reliable using quasi-automatic 3D reconstruction methods with biplanar X-rays. MethodsConsecutive children with thoracic idiopathic scoliosis who underwent spinal fusion surgery and biplanar pre and postoperative X-rays in free-standing position were included prospectively. Quasi-automatic computed 3D reconstructions of the spine were done using a previously validated technique and allowed the automatic computation of geometrical spinopelvic parameters including OD-pelvis, linear, and developed T1-T12 and T1-L5 lengths. ResultsThirty patients with scoliosis were included, and 240 reconstructions were performed (2 operators x2 repetitions x30 patients pre and postoperative). The main thoracic Cobb angle, T1-T12, T1-L5 linear and developed distance, OD-pelvis were significantly improved (p < 0.001). The gain of the main thoracic Cobb angle (31.6°;SD = 9°) was correlated to the gain of the linear distance T1-T12 (15.3 mm;SD=7.3 mm)(rho = 0.76;p < 0.0001) and T1-L5 (24.7 mm;SD = 8 mm)(rho = 0.64;p < 0.0001). The postoperative change of developed length between T1-L5 represented 41% of the gain in linear distance between the same vertebrae. Similarly, the gain of T1-T12 developed length was 50% of linear T1-T12 height gain. Both differences were significant (p = 0.01). Absolute bias using Bland & Altman plots was lower than 1 mm for linear distance (0.1%) and lower than 2 mm (0.3%) for developed distance. ConclusionThe gain in spinal length is correlated to the thoracic Cobb angle correction in the surgical treatment of idiopathic thoracic scoliosis. The new significant finding is that the developed spinal height gain represented approximately a little less than 50% of the linear spinal height gain and these parameters were reliable from a 3D quasi-automatic reconstruction of biplanar X-ray.