Automatic reduction planning of pelvic fracture based on symmetry

Pelvic fracture is known to be one of the most difficult injuries. The first and foremost step for pelvic fracture reduction is structural recovery. However, due to lack of data on healthy pelvis before fracture, the result of fracture reduction is entirely based on the experience of surgeons. This study proposed a novel method for automatic reduction planning based on the symmetry of the pelvis (pelvic symmetry reduction, PSR), which could provide a template for the pelvic reduction surgery as well as related surgical robot. A symmetrical area (ROI) without fracture fragments was extracted from the fractured pelvis through multiple SP mirroring, ICP registration and Boolean operation, so as to acquire the target position for reduction. Fifteen healthy pelvic models and three fractured pelvises were acquired to evaluate the accuracy and validity of the PSR method separately. The mean error of anatomical landmarks is 2.94 mm (translational error) and 1.08°, 0.67°, and 1.96° (rotational error on X, Y and Z axes, respectively). The average global distance error is below 4 mm except for a model with poor symmetry. The results of the actual fractured pelvis were recognised by clinicians. These results demonstrate the feasibility of the PSR method for restoring anatomical structure of the pelvis and achieve higher reduction accuracy than previous approach.