|dc.description.abstract||We are witnessing increasing applications of information technologies in all branches of medicine. The introduction of modern technologies in the field of surgery and orthopedics in everyday practice has allowed physicians to perform diagnostic, preventive and therapeutic activities in the best way. The success of these activities, in addition to the available devices, depends on the experience of doctors and the available information about the patient.
Incomplete and inaccurate information on morphology, geometry and the structure of human bones, that are necessary for the reconstruction of bone geometric models, have encouraged the development of new and improvement of existent methods of creating geometrical models of human bones based on predictions. Most of the existent methods for geometrical modeling of the human bones are based on input data (input data of a patient’s bone) and the application of statistical methods for processing these data. As a result of the process, geometrical models of human bones are obtained which can be used for different preclinical and clinical needs.
The main goal of this doctoral dissertation has been to development of a method which would enable of creation of a 3D geometrical model of human bones, in this case the human mandible (complete bone, as well as of the missing bone parts) based on complete and / or incomplete entrance data of patients bones. The model thus created would be used in preoperative preparation, simulation of surgical interventions, production of osteofixation material and personalized bone implants, enabling doctors to make appropriate decisions in everyday clinical practice.
A special emphasis is placed on improving the existent method of reverse engineering - Method of Anatomical Features (MAF), which allows the creation of geometrical models of human bones in cases where the input data of patient’s bone are complete or not complete. This method served as the initial basis for further research analyzes aimed at solving the problem of the prediction of the human bone geometry.
In accordance with the defined goals, a new approach for the geometrical modeling of the human mandible based on the prediction of human bone geometry using artificial neural network techniques was realized. The modeling methodology is based on the created mathematical model where input data represent the values of the specific parameters acquired from the medical images, while the output data represent the values of the anatomical entities (coordinates of anatomical points on the surface on the bone). By establishing mathematical relations between specific parameters and anatomical
entities, a precise description of the geometrical entities of human bones is enabled. As result of the applied approach the new algorithm for creation a 3D parametric model of the human mandible is formed.
The modeling methodology presented in this dissertation was implemented through the process of creation the complete bone of the human mandible. In order to verify the method, the obtained results were compared with the results obtained by other statistical methods. Verification of the results was carried out through a comparative analysis of geometry and deviation between initial and constructed models.
An additional contribution to the research is the application of the developed 3D parametric model of the human mandible in characteristic cases which can appear in clinical practice, case of the creation of the missing parts of the bone structure and case of the elimination of bone deformities.
The research results presented in this paper indicate that the new approach in geometrical modeling of human mandible, gives better and more precise the resulting 3D models of the complete bone, as well as every anatomical region individually, even in the cases where the input data of a patient’s bone are not complete. Based on everything stated above, the new approach provides a significant scientific contribution to improvement of existent method used in reverse engineering of human mandible.
Research contributions are reflected in: a more precise and accurate creation of the 3D parametric model of the human mandible, on a better qualitative and quantitative assessment of geometrical model parameters, and the development of a new prediction method.||en