Hoen-Oh Shin MD, Presenter: Nothing to Disclose

Essig Harald MD, Abstract Co-Author: Nothing to Disclose

Michael Jagodzinski MD, Abstract Co-Author: Nothing to Disclose

Frank K. Wacker MD, Abstract Co-Author: Research grant, Siemens AG

Automatic 3D modeling of complex bone fractures and defects as well as the automatic design of individual implants are feasible using the proposed computer algorithm. It is highly adjustable to different anatomic regions.

Preoperative planning for computer-assisted surgery (CAS) of complex bone trauma requires accurate 3D models with reconstruction of bone defects using high resolution CT and MRI data. For one-sided defects of anatomical structures with mirror symmetry such as the scull, patient specific computer models can be generated by mirroring the unaffected healthy half of the structure to the pathological side along the centerline. However, neither complex fractures crossing the centerline or body parts with asymmetric geometry can be modeled with this technique. The proposed software-tool allows for a fully automatic 3D modeling of complex bone fractures or defects utilizing a reference data base that are automatically adapted to the patient-specific data by means of a computer algorithm.

3D models were derived from Multi-detector-row CT data from 9 patients with complex maxillofacial fractures involving both side of the skull and 2 patients prior to knee arthroplasty. Dislocated fractures and bone defects were modeled by the best matching subject of a reference data base by means of a registration algorithm. Generation of virtual anatomic models was successful in all 11 patients. The maximum deviation between model data and patient data was 2.5 mm. Computing time for registration was 3 hours and 26 minutes on average (Mac pro, 2 dual-core CPUs, 32 GB RAM).  

Software for patient specific 3d modeling and design of implants is often complex and restricted to a specific anatomic region and pathology – if available at all. The proposed technique is highly adjustable to different anatomic regions and pathologies only requiring a small reference data base of the anatomic region of interest. Depending on the size of the imaging data, computing time can be high taking several hours on typical CT data sets of several hundred slices using actual workstations. However, no manual interaction is required making the use of the algorithm feasible in a clinical setting.

Shin, H, Harald, E, Jagodzinski, M, Wacker, F, Automatic Reconstruction of Bone Defects for Preoperative Planning and Modeling of Implants.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11014601.html