Anteroposterior (AP) pelvic radiographs are the standard tool used for pre-operative planning and post-operative evaluation during total hip arthroplasty (THA). The accuracy of this imaging modality is, however, limited by errors in pelvic orientation and image distortion. Pelvic obliquity is corrected for by orienting measurements to a reference line such as the interteardrop line or the interischial line, while several methods for correcting for pelvic tilt have been suggested, with varying levels of success. To date, no reliable method for correcting for pelvic rotation on pelvic imaging is available. The purpose of this study was to evaluate a novel method for correcting pelvic rotation on a standard anteroposterior (AP) radiographs. Computed tomography (CT) scans from 10 male cadavers and 10 female THA patients were segmented using 3D Slicer and used to create 3D renderings for each pelvis. Synthetic AP radiographs were subsequently created from the 3D renderings, using XRaySim. For each pelvis, images representing pelvic rotation of 30° left to 30° right, at 5° increments were created. Four unique parameters based on pelvic landmarks were used to develop the correction method: i) the horizontal distance from the upper edge of the pubic symphysis to the sacroiliac joint midline (PSSI), ii) the ratio of the horizontal distances from the upper edge of the pubic symphysis to the outer lateral border of both obturator foramina (PSOF), iii) the width ratio of the obturator foramina (OFW) and iv) the ratio of the horizontal distance from each anterior superior iliac spine to the sacroiliac joint midline (ASISSI). The relationships between the chosen parameters and pelvic rotation were investigated using a series of 260 (13 per pelvis) synthetic AP radiographs. Male and female correction equations were generated from the observed relationships. Validation of the equations was done using a different set of 50 synthetic radiographs with known degrees of rotation. In males, the PSSI parameter was most reliable in measuring pelvic rotation. In females, PSOF was most reliable. A high correlation was noted between calculated and true rotation in both males and females (r=0.99 male, r=0.98 female). The mean difference from the male calculated rotation and true rotation value was 0.02°±1.8° while the mean difference from the female calculated rotation and true rotation value was −0.01°±1.5°. Our correction method for pelvic rotation using four pelvic parameters provides a reliable method for correcting pelvic rotation on AP radiographs.

For any figures or tables, please contact authors directly.