Client: Orthopaedic Biomechanics Lab – Sunnybrook Research Institute
Goals: Develop a free, open-source 3D virtual simulator intended to provide surgeons and residents with an accessible means to plan and practice pedicle screw insertion.
Made with: Python, 3D Slicer, Maxon Cinema 4D, Adobe After Effects
Pedicle screws are often used in spinal fusion surgery as a means of gripping a spinal segment. However, pedicle screw insertion is a challenging procedure, with complication rates ranging from 1% to 54%. Although preoperative planning can be carried out by measuring distances and angles using computed tomography (CT) and magnetic resonance imaging (MRI), when in the operating room, surgeons insert screws based on feel and alignment with spatial landmarks. Surgical residents receive hands-on training using cadaveric specimens, however these opportunities are largely dependent on the availability of time and lab space.
Surgical simulators offer residents the chance to practice pedicle screw insertion in a virtual operating room. However, most of these simulators run on expensive commercial platforms with single-seat licenses. Described here, is the development of a free, open-source 3D virtual simulator intended to provide surgeons and residents with an accessible means to plan and practice pedicle screw insertion.
The simulator was built as a Python-scripted module for the open-source image analysis software package, 3D Slicer. The simulator allows users to import patient specific CT or MRI scans as DICOM data sets, and can be fully integrated with Picture Archiving and Communication Servers (PACS). Users can then use 2D orthographic views along with 3D renderings of a patient’s spine to make measurements in preparation for surgery. Using Maxon Cinema 4D and Adobe After Effects, instructional videos were made to assist residents with finding key anatomical landmarks. These videos are accessible directly within the simulator. Users are then immersed in an interactive environment, in which they can manipulate 3D models of pedicle screws and insert them into the model of the patient’s spine. Screw placement can then be evaluated by calculating the percentage of cortical and cancellous bone in contact with the surface of each screw, and displayed graphically. Additionally, the position of each screw model is compared with the patient’s CT data to map Hounsfield unit pixel intensities directly onto the surface of the screw model. This effectively allows users to easily identify areas of screws that have breached the bone surface.