Grant: Collaborative Health Research Projects (NSERC partnered)

Tremendous advances have occurred in fracture management, due in part to new implants that internally fixate many fractures with minimally invasive techniques. For example, complex femoral fractures are now treated by sophisticated 'locking' nails that require only a few small incisions to insert, giving better outcomes. Despite these advances, articular fractures - which occur near joints - are still challenging even when using specialized locking plates, in part because of the difficulty in achieving anatomic reduction of the joint surface (articular surface fractures that heal with 2 mm or more of residual displacement are at high risk of causing arthritis). To better exploit these new fixation devices we need better minimally invasive surgical techniques, but there currently is no efficient intraoperative means of visualizing bone fragments in 3D. In particular, complex multi-fragment articular fractures are routinely reduced using surgical exposures that may increase scarring, devascularize bone or joint surfaces, and dislodge previously undisplaced bone fragments. Less invasive techniques often involve liberal usage of conventional 2D X-ray fluoroscopy that often provides suboptimal images, and that also exposes the patient and surgical team to significant doses of radiation. Image-guided surgery uses preoperative and intraoperative images to aid the surgeon in performing tasks for which direct vision is impossible or inappropriate. Current commercial navigation systems use 2D fluoroscopy, and a tracking device affixed to the fracture fragment, to estimate a 3D registration transformation between the segmented images and the patient. Such a technique cannot be used for small fracture fragments - particularly articular fractures - because: (1) fixation of a tracking device could damage articular surfaces; (2) the registration of 2D X-ray images to the 3D anatomy is a challenging task under operating room conditions; (3) no automated segmentation algorithms exist for fractures, and (4) manual segmentation of imaging data is time and labour intensive, requiring skilled technical staff. The associated costs significantly limit the clinical utility and availability of current navigation interfaces for fracture reduction