A computer aided measurement method for unstable pelvic fractures based on standardized radiographs

BACKGROUND To set up a method for measuring radiographic displacement of unstable pelvic ring fractures based on standardized X-ray images and then test its reliability and validity using a software-based measurement technique. METHODS Twenty-five patients that were diagnosed as AO/OTA type B or C pelvic fractures with unilateral pelvis fractured and dislocated were eligible for inclusion by a review of medical records in our clinical centre. Based on the input pelvic preoperative CT data, the standardized X-ray images, including inlet, outlet, and anterior-posterior (AP) radiographs, were simulated using Armira software (Visage Imaging GmbH, Berlin, Germany). After representative anatomic landmarks were marked on the standardized X-ray images, the 2-dimensional (2D) coordinates of these points could be revealed in Digimizer software (Model: Mitutoyo Corp., Tokyo, Japan). Subsequently, we developed a formula that indicated the translational and rotational displacement patterns of the injured hemipelvis. Five separate observers calculated the displacement outcomes using the established formula and determined the rotational patterns using a 3D-CT model based on their overall impression. We performed 3D reconstruction of all the fractured pelvises using Mimics (Materialise, Haasrode, Belgium) and determined the translational and rotational displacement using 3-matic suite. The interobserver reliability of the new method was assessed by comparing the continuous measure and categorical outcomes using intraclass correlation coefficient (ICC) and kappa statistic, respectively. RESULT The interobserver reliability of the new method for translational and rotational measurement was high, with both ICCs above 0.9. Rotational outcome assessed by the new method was the same as that concluded by 3-matic software. The agreement for rotational outcome among orthopaedic surgeons based on overall impression was poor (kappa statistic, 0.250 to 0.426). Compared with the 3D reconstruction outcome, the interobserver reliability of the formula method for translational and rotational measures was perfect with both ICCs more than 0.9. CONCLUSIONS The new method for measuring displacement using a formula was reliable, and could minimise the measurement errors and maximise the precision of pelvic fracture description. Furthermore, this study was useful for standardising the operative plan and establishing a theoretical basis for robot-assisted pelvic fracture surgery based on 2-D radiographs.