Analysis and simulation of the distal forearm stability during pro-supination

Introduction: The radioulnar joint is one of two joints of the forearm bones radius and ulna. It is anatomically located at the wrist. The bone surfaces enable forearm motion (i.e., pro-supination) in an extensive range while the stability is primarily maintained by a complex system of ligaments connecting the bones. Consequently, soft tissue injury can cause chronic joint instability, pain, or functional disability. In these patients, operative treatment by anatomical reconstruction of the injured ligament using a tendon graft is indicated. Bone realignment by corrective surgery is additionally mandatory if a ligament tear is accompanied by bone deformity. The surgical outcome, however, remains unpredictable in many cases because the effect of the surgery on the forearm mobility cannot be accurately assessed before a surgery. The reason for this is that state-of-the-art computer simulations do not take into account any soft tissue structures. For corrective surgeries of the forearm bones, computer assisted planning relies only on a comparison to the healthy bone of the opposite side. These approaches can obviously not be applied to pathologies where soft tissue components play a key role. Methods are therefore needed to allow the patient-specific modelling and subsequent simulation of soft tissue structures of the forearm during pro-supination motion. Goals of this project: Aim is to develop a patient-specific hard- and soft-tissue model for simulating not only the healthy but also the pathological forearm motion with respect to functional disability and instability. Besides improving the diagnosis of soft tissue associated injuries before an intervention, such as for distal radius instability, the ultimate goal is to predict the surgical outcome by analyzing the forearm motion before and after simulated surgery. This would enable the surgeon to select the optimal treatment specific to the pathology and biomechanics of the patient. Project design: The project is divided into three main parts, namely anatomical modelling, simulation, and a clinically-focused part. The key idea for understanding and learning the motion pattern will be the acquisition of static and dynamic in-vivo data of the pathological and contralateral healthy forearm. To do so, a clinical trial with patients suffering from distal radioulnar joint instability will be performed. Fluoroscopy tracking combined with 2D-3D registration will be employed for collecting accurate kinematic data. The modelling part will focus on the development of methods for generating a 3-dimensional representation of the relevant patient anatomy (i.e., bones with ligaments) from computed tomography and magnetic resonance images in an automatic fashion. The data obtained from the clinical trial will serve as a basis for the development of the computer model of the (pathological) forearm motion. The simulation model will be embedded into an existing preoperative planning framework for predicting functional improvement. Lastly, the methods will be validated in cadaver experiments. Significance of the planned work: More accurate planning before a surgery will result in a more precise surgical procedure and, consequently, in a better surgical outcome. The resulting improvement in joint mobility and pain reduction contributes to the quality of life of the patients and reduces occupational invalidity. The approach combines patient-specific anatomical modelling of soft tissue structures with motion acquisition and simulation in a flexible way. Therefore, the concept may be applied to other joints of the extremities for improving diagnosis of complex functional pathologies. The novel segmentation algorithms will boost 3D-based preoperative diagnosis of orthopedic surgeries in general because automated segmentation of complex ligamentous structures from medical image data is currently not feasible in daily clinical routine.

The aim of the project was to develop a patient-specific bone and soft tissue movement model that simulates not only healthy movement but also motion limitations due to restriction or instability. In addition to improving diagnosis before surgery, the motion model should also be used effectively to predict the outcome of surgery. The analysis of the motion before and after simulated surgery should enable the optimal treatment method to be selected specifically for the patient and his or her situation. The exact segmentation of the connective tissue is still a difficult task, which has to be solved for the generation of geometric models for biomechanical calculations. Alternatively, it is possible to predict the ligament insertion sites and then approximate geometries based on anatomical knowledge and morphological studies. For this purpose, the project has developed an integrated framework for the automatic modelling of human musculoskeletal ligaments. Statistical shape modeling was combined with geometric algorithms for automatic identification of insertion sites. Based on these, geometric surface and volume meshes were generated. For the adaptation to the anatomy of the forearm, insertion sites of ligaments were determined in the statistical model according to an anatomical approach from previously published papers. For evaluation, these were compared with data from a cadaver study with five forearms or a total of 15 ligaments. The system allowed the creation of 3D models that approximated the true shape of the ligaments. However, it was found that the statistical model was not always reliable. The average mean square error as well as the Hausdorff distances of the meshes sometimes increased by more than one order of magnitude. Instead, using the data from the cadaver study, the average mean square error was 0.59 mm and the Hausdorff mean distance was less than 7 mm. In general, the approach to automatically generate ligament geometries from insertion points proved to be practicable; however, the use of those points from a statistical model was too imprecise.