Novel method for selecting slices of the same cross-sectional view from digital tomosynthesis for monitoring posterior spinal instrumentation.

Abstract Background A method to evaluate pedicle screw loosening on digital tomosynthesis images is yet to be established owing to lack of methods for selecting slices of the same cross-sectional view. We aimed to develop an objective method for selecting slices of the same cross-sectional view on digital tomosynthesis images. Methods First, an objective method of pixel selection was developed by measuring the size of glass disk and titanium alloy screw on digital tomosynthesis images followed by comparison with the actual sizes. Second, a method for selecting slices of the same cross-sectional view was explored on a bone model with posterior spinal instrumentation using the screw centerline and rod curvature as indicators of the same cross section. The angle between the screw centerline and rod was calculated to verify the accuracy in obtaining the same cross-sectional view. Third, the method for selecting slices of the same cross-sectional view was applied to six patients after posterior lumbar spinal instrumentation. Results The pixel selection method enabled objective determination of a pixel on the peripheral lines of objects with an error as low as 200 μm in distance measurements on titanium alloy and glass. The mean differences of rod-screw angles between two slices were less than 1° and were not statistically significant in the bone model and patient images. Conclusion A method for selecting slices of the same cross-sectional view on digital tomosynthesis images was successfully developed. This method can enable objective and quantitative evaluations of pedicle screw loosening.