Study of a bone segmentation and 3D modeling workflow for diagnosis and planning

This PhD thesis presents the development and evaluation of a comprehensive workflow for bone segmentation and 3D modeling aimed at diagnostic support and preoperative planning. The work is divided into two main areas: the usability assessment of a multimodal biomedical imaging platform ("All-in-One") and the study of image processing methods for accurate bone structure extraction. The usability study was carried out through the design of a formal evaluation protocol based on international standards, involving real users and scenarios. On the technical side, the thesis explores both traditional and deep learning-based approaches for bone segmentation, including the implementation of a customized U-Net architecture optimized for CBCT images. A dedicated tool, the Single Bone Modeler (SBM), was developed to extract individual bones from complex anatomical regions and generate high-quality 3D models. This tool will be integrated into the All-in-One platform. The proposed workflow has demonstrated high potential for enhancing clinical decision-making, and future work will focus on the integration of soft tissue analysis and the use of 3D printing technologies for surgical planning.