Spinal Palpation Error and Its Impact on Skin Marker-Based Spinal Alignment Measurement in Adult Spinal Deformity

Abstract

Spinal alignment measurement in spinal deformity research has recently shifted from using mainly two-dimensional static radiography toward skin marker-based motion capture approaches, allowing three-dimensional (3D) assessments during dynamic conditions. The validity and accuracy of such skin marker-based methods is highly depending on correct marker placement. In this study we quantified, for the first time, the 3D spinal palpation error in adult spinal deformity (ASD) and compared it to the error in healthy spines. Secondly, the impact of incorrect marker placement on the accuracy of marker-based spinal alignment measurement was investigated. 3D, mediolateral and inferosuperior palpation errors for thoracolumbar and lumbar vertebral levels were measured on biplanar images by extracting 3D positions of skin-mounted markers and their corresponding anatomical landmarks in 20 ASD and 10 healthy control subjects. Relationships were investigated between palpation error and radiographic spinal alignment (lordosis and scoliosis), as well as body morphology [BMI and soft tissue (ST) thickness]. Marker-based spinal alignment was measured using a previously validated method, in which a polynomial is fit through the marker positions of a motion trial and which allows for radiograph-based marker position correction. To assess the impact of palpation error on spinal alignment measurement, the agreement was investigated between lordosis and scoliosis measured by a polynomial fit through, respectively, (1) the uncorrected marker positions, (2) the palpation error-corrected (optimal) marker positions, and (3) the anatomically corrected marker positions (toward the vertebral body), and their radiographic equivalents expressed as Cobb angles (ground truth), using Spearman correlations and root mean square errors (RMSE). The results of this study showed that, although overall accuracy of spinal level identification was similar across groups, mediolateral palpation was less accurate in the ASD group (ASDmean: 6.8 mm; Controlmean: 2.5 mm; p = 0.002). Significant correlations with palpation error indicated that determining factors for marker misplacement were spinal malalignment, in particular scoliotic deformity (r = 0.77; p < 0.001), in the ASD group and body morphology [i.e., increased BMI (rs = 0.78; p = 0.008) and ST thickness (rs = 0.66; p = 0.038)] in healthy spines. Improved spinal alignment measurements after palpation error correction, shows the need for radiograph-based marker correction methods, and therefore, should be considered when interpreting spinal kinematics.