Three-Dimensional Computer-Assisted Methods for the Automation and Optimization of Preoperative Planning of Forearm Orthopaedic Surgeries

The outcome of complex orthopaedic surgeries highly depends on three main factors: the accurate assessment of the pathological condition, the careful preoperative planning of the surgical steps, and the subsequent implementation of the preoperative planning by corresponding intraoperative guidance. The restoration of the pretraumatic anatomy is one of the ultimate goals in reconstructive surgery of (posttraumatic) bone deformities. The preferred surgical treatment for reconstructive surgeries is a procedure known as corrective osteotomy, where the malunited bone is cut into one or more fragments and reduced to the correct anatomical position represented by a reconstruction template. High accuracy and precision are desired during the assessment, planning and intraoperative navigation of these procedures, as a deviation from the preoperative plan may result in unsatisfactory clinical outcomes. Thereby, three dimensional (3D) computer-assisted (CA) methods have become the state-of-the-art for the assessment of bone deformity and the generation of the corresponding preoperative planning of corrective osteotomies, which are later translated into the intraoperative situation by means of patient-specific navigation instruments. The benefits of newly available technologies for the 3D CA planning in orthopaedics comes at the cost of an increase in complexity of the processes involved in the generation of an optimal preoperative plan, requiring a vast clinical experience and high level of technical expertise on simulation and graphical tools. Moreover, the corresponding steps for the generation of a 3D preoperative plan are highly interdependent and often challenging to optimize manually, generating long planning iterations and incurring undesired clinical cost. Hence, there is a need for automation of the preoperative planning process, in order to reduce costs and deduct valuable time from the clinical setting. Additionally, the automation of the preoperative process would allow repeatability of the solutions and enable to systematically investigate the most optimal preoperative plan according to given clinical goals. Hence, the primary aim of this work is the development of a novel computer algorithm framework, capable of generating a preoperative plan for corrective osteotomies, in a fully automatic fashion. Furthermore, the increasing interest of clinicians for a more comprehensive analysis of complex pathologies often requires dynamic models able to simulate a larger range of interventions. Nevertheless, current CA preoperative methods are limited to simple bone procedures. One missing piece to enable more complex planning simulations is the inclusion of soft tissue influence into the preoperative process. Therefore, our second objective is to address the missing influence of the soft tissue in the preoperative planning by providing the basis for the integration of the interosseous membrane (IOM) into the forearm motion analysis. In a first step, 3D automated methods are developed for the assessment of the distal radioulnar joint (DRUJ) morphology based on bony landmarks, incorporating subject-specific cartilage surface orientation. Gold-standard radiological measurements on a large consecutive dataset of healthy DRUJ are compared against the measurements obtained through the developed methods. Afterwards, the ability of the developed methods to quantify the 3D DRUJ morphology is evaluated. These methods enable the translation of standardized measurements used in the field of hand surgery into 3D space. The main part of the thesis focuses on the development of an automatic optimization framework for the generation of preoperative planning solutions of the forearm. The framework is based on a genetic algorithm optimization approach, capable of handling multiple clinical objectives and able to provide surgeons with complete preoperative planning solutions. The approach provides the osteotomy cut, the required bone reduction and the optimal positions for the osteosynthesis implant and fixation screws. Automatic methods for deformity assessment of long-bone deformities are developed and clinical objectives are translated into the optimization framework by means of tailored fitness functions. The clinical feasibility of the approach is evaluated on consecutive series of radius osteotomies by comparing solutions generated by the optimization algorithm against gold-standard solutions previously obtained by the surgeons. In addition, the technical capabilities of the framework are evaluated using standard clinical metrics. This optimization framework is capable of generating ready-to-use preoperative solutions without requiring further manual adjustments, significantly reducing planning times and consequently associated costs. Moreover, we show that the algorithm is capable of outperforming the state-of-the-art method by generating solutions that are not possible to obtain through a manual optimization technique. The remaining parts of this thesis aim to contribute to the inclusion of soft-tissue influence into the preoperative planning in order to tackle the problem of missing simulations able to cover a larger range of complex pathologies. Thereby, we have generated 3D morphological data of the interosseous membrane of the forearm, and provide a first insight into the feasibility of soft-tissue integration into forearm motion analysis. Methods are developed for the calculation of common morphological features on 3D models of the IOM, generated from cadaveric forearms through a combination of different image modalities. Afterwards, tensile properties of the individual ligaments of the forearm are obtained, which are currently missing and are a basic prerequisite for the construction of more accurate forearm simulation models. Finally, a forearm simulation prototype is developed including the individual ligaments of the IOM and the forearm bones. This kinematic simulation provides insights on the influence of the individual IOM ligaments on the forearm motion