Assessment of scoliotic deformity using spinous processes: comparison of different analysis methods of an...

The purpose of this study was to evaluate the performance of 5 analysis methods in quantifying scoliotic deformity, using the spatial positions of SP tips acquired by a custom-developed ultrasound-based system, with different curve fitting methods and angle metrics in terms of their correlation with Cobb angle, test-retest reliability, vulnerability to digitization errors, and accuracy of identifying end vertebrae and convexity direction.

Three spinal column dry bone specimens were randomly configured to 30 different scoliotic deformities. Raw spatial data of the SP tips were processed by the following 3 methods: (1) fifth-order polynomial fitting, (2) locally weighted polynomial regression (LOESS) with smoothing parameter (α) = .25, and (3) LOESS with α = .4. Angle between the 2 tangents along the spinal curve with the most positive and negative slopes (ie, posterior deformity angle) and summation of the angles formed by every 2 lines joining 3 neighboring SPs between the end vertebrae (ie, accumulating angle) were computed to quantify scoliotic deformity. Their performances were compared in terms of their correlation with Cobb angle, test-retest reliability, vulnerability to digitization errors, and accuracy of identifying end vertebrae.

Posterior deformity angle calculated from the spinal curve constructed by LOESS with α = .4 excelled in every aspect of the comparison (ie, Cobb angle, test-retest reliability, vulnerability to digitization errors, and accuracy of identifying end vertebrae and convexity direction), making it the method of choice of those tested for processing the spatial data of the SP tips in this ultrasonography study using dry bone specimens.

The ultrasound-based system and the LOESS (0.4)-posterior deformity angle method developed for this study offer a viable technology for quantifying scoliotic deformity in a reliable and radiation-free manner. However, further validation using scoliosis subjects is needed before they can be used to quantify spinal deformity in the clinical setting.