A 3D environment for surgical planning and simulation

The use of Computed Tomography (CT) images and their three-dimensional (3D) reconstruction has spread in the last decade for implantology and surgery. A common use of acquired CT datasets is to be handled by dedicated software that provide a work context to accomplish preoperative planning upon. These software are able to exploit image processing techniques and computer graphics to provide fundamental information needed to work in safety, in order to minimize the surgeon possible error during the surgical operation. However, most of them carry on lacks and flaws, that compromise the precision and additional safety that their use should provide. The research accomplished during my PhD career has concerned the development of an optimized software for surgical preoperative planning. With this purpose, the state of the art has been analyzed, and main deficiencies have been identified. Then, in order to produce practical solutions, those lacks and defects have been contextualized in a medical field in particular: it has been opted for oral implantology, due to the available support of a pool of implantologists. It has emerged that most software systems for oral implantology, that are based on a multi-view approach, often accompanied with a 3D rendered model, are affected by the following problems: unreliability of measurements computed upon misleading views (panoramic one), as well as a not optimized use of the 3D environment, significant planning errors implied by the software work context (incorrect cross-sectional planes), and absence of automatic recognition of fundamental anatomies (as the mandibular canal). Thus, it has been defined a fully 3D approach, and a planning software system in particular, where image processing and computer graphic techniques have been used to create a smooth and user-friendly completely-3D environment to work upon for oral implant planning and simulation. Interpolation of the axial slices is used to produce a continuous radiographic volume and to get an isotropic voxel, in order to achieve a correct work context. Freedom of choosing, arbitrarily, during the planning phase, the best cross-sectional plane for achieving correct measurements is obtained through interpolation and texture generation. Correct orientation of the planned implants is also easily computed, by exploiting a radiological mask with radio-opaque markers, worn by the patient during the CT scan, and reconstructing the cross-sectional images along the preferred directions. The mandibular canal is automatically recognised through an adaptive surface-extracting statistical-segmentation based algorithm developed on purpose. Then, aiming at completing the overall approach, interfacing between the software and an anthropomorphic robot, in order to being able to transfer the planning on a surgical guide, has been achieved through proper coordinates change and exploiting a physical reference frame in the radiological stent. Finally, every software feature has been evaluated and validated, statistically or clinically, and it has resulted that the precision achieved outperforms the one in literature.