Knowledge-based design of lower limb prosthesis

During last years the development of many products has been improved by the introduction of computer-aided tools reducing the time and costs of the full development process, and allowing to evaluate in a faster and cheaper way different variants of the same product. Besides, a great attention has been put on custom-fit products characterized by a close interaction with the human body or part of it. Innovative computer-aided tools can help to realise custom-fit products with a strict interaction with human body and definitely improve people’s quality of life, in particular of persons with disabilities. The work presented in this thesis refers to this context and to a specific custom-fit product: the lower limb prosthesis. Main objective has been to realize an innovative knowledge-based framework, centred on the virtual models of the patient’s body, which can guide and support the user during all the steps of the prosthesis design, suggesting rules and procedures for each task. Research activities have been organized into four main phases as follows. First it has been studied the State of the Art of prosthesis, the modular ones, and of ICT tools for socket design available on the market. A modular prosthesis is mainly composed by commercial components, except for the socket that is highly customised and manufactured around the patient stump. This component is the interface between the stump and the mechanical part of the prosthesis and requires high level of customization in order to satisfy functional and comfort requirements. Regarding known ICT tools we have verified that they can support some specific tasks of the product development process, but they do not offer any kind of assistance to the prosthetists. In fact, design process decisions and actions are taken on the base of technicians’ experience and personal skills. Therefore, it is strategic to integrate within such systems specific domain knowledge, in order to obtain a valid and high quality final product, and to develop for this reason a design framework which can assist the technician during all the process. The second phase consisted in acquiring the knowledge related to the product and the traditional process to manufacture modular lower limb prosthesis, reviewing specific literature and scientific publications and, above all, interviewing qualified orthopaedic technicians. This activity has been carried out in an orthopaedic laboratory, participating and following personally all the stages of the prosthesis manufacturing processes. All information has been formalized with IDEF0 diagrams, deriving all the implicit design rules and procedures. It has been highlighted that all the product and process knowledge is strictly correlated to a specific set of parameters, which guides the whole prosthesis design process: the patient’s characteristics. In particular, these data are necessary to select the appropriate standard components, to model a functional socket in correlation to the patient’s anatomy, and to correctly size the final prosthesis assembly. From this phase we have extrapolated: • Patient parameters guiding the whole process; • Selection procedures to choose the most appropriate standard components; • Rules to correctly size the prosthesis; • Rules and procedures to create the 3D model of the socket. In the third phase, on the basis of previous analysis, we have reengineered the prosthesis design process and developed a knowledge-base design framework, which guides the technicians step by step providing for each activity specific knowledge and rules (e.g. dimensioning or selection rules for standard parts). The proposed framework is centred on the digital model of the amputee and directly manages the experts’ knowledge in order to guarantee a product of high quality. Each activity is supported in direct way by the management of specific domain knowledge through virtual assistants, which provide procedures and/or suggestions to perform best choices and, when possible, execute them automatically (e.g., choosing components and materials, or sizing parts). It integrates ad hoc tools for domain technical knowledge management both of product and process, virtual modelling of components both standard (e.g. pylons and tubes, prosthetic feet, etc.) and custom-fit, and tools for behaviour simulation (e.g. by means of FE and multibody techniques) to investigate component-human body interaction. In particular, the work developed in this thesis has been focused on modelling issues. The framework has been implemented using a commercial KBE system (Ruledesigner® Configurator). A commercial 3D CAD system (Solid Edge Siemens PLM Software) has been adopted to create a library of 3D parametric models to represent standard components and for final prosthesis assembly. The rules and procedures extrapolated for socket design have been embedded within a module, named Socket Modelling Assistant (SMA), specifically developed by V&K Research Group (University of Bergamo) for socket virtual modelling. SMA provides a set of virtual tools that permit to emulate the tasks traditionally performed by the technicians. Last phase concerned the validation and experimentation of the framework. First we have tested and refined the procedures for standard components selection in collaboration with the involved orthopaedic laboratory. By means of the framework, we have set up prosthesis configuration for ten patients, five transfemoral and five transtibial and compared with those ones proposed by the prosthetics. Results have been considered satisfactory for most of the cases. It has been appreciated the possibility to generate rapidly alternative configurations. Secondly, procedures and rules for socket modelling have been tested creating the 3D model of the socket for two patients, one transfemoral and one transtibial. Together with the technicians, we mainly tested the efficacy of the shape manipulation tools that virtually emulate traditional procedures. Preliminary results have been considered positively by the laboratory technical staff but they envisaged the need of some modifications to make mentioned tools easier to use, especially, by a non computer-skilled end-user. Finally, for a better evaluation of socket model, we need to compare it with the real one manually realised.