Final Report Summary - MULTISCALEHUMAN (Multi-scale Biological Modalities for Physiological Human Articulation) The Multi-scale Biological modalities for physiological human Articulation (MultiScaleHuman) project is supported by 7 institutes across 4 countries (Switzerland, Germany, Italy and Portugal) and is coordinated by Prof. Dr. Nadia Magnenat-Thalmann, director of MIRALab at the University of Geneva. The main goal of MultiScaleHuman is to develop, by training early stage (ESR) and experienced (ER) researchers, a multi-scale biological data visualization and knowledge management system. Research and training activities have provided an improved understanding, diagnosis and treatment of musculoskeletal diseases (MSD) in physiological human articulations. MultiScaleHuman has exploited advances in multi-scale biological modalities and their integration by addressing five core biological scales: Molecular, Cellular, Tissue, Organ and Behaviour scales. As such, a multi-layered understanding and proof of concept of the functional links between scales have been investigated. A detailed description and contact information can be found on the project website: http://multiscalehuman.miralab.ch/. Based on the solid foundation laid in the first period, the MSH project has generated high quality results and many publications in established journals and conferences. The detailed schedule of secondments for the fellows has led to numerous collaborations that highlight and support the multi-scale approach of this project.HUGE has focused on morphological and biological modeling of muscular activity of lower limbs obtained from multimodality data obtained from PET-MR imaging. HUGE has defined the MRI and PET acquisition protocols together with MIRALab and has performed a study with five subjects. All partners have benefited from the resulting MR images. Dynamic acquisitions were used to model the muscle deformation during exercise. Specific biological modeling of metabolic tracers and quantitative image analysis tools were developed based on laboratory experiments and evaluated to have a first understanding of physiological and mechanical behaviour of muscles and articulations.UMINHO has developed a 3D model of human OA cartilage, which enables the study of the disease progression in vitro, on multiple scales (molecular, cellular, tissue). The concept of multiscale OA modelling in vitro, and the obtained data was provided to the partners in the project (Welfenlab, CNR-IMATI) who used it to build an integrated visualisation system. The tissue was maintained in vitro by using bioreactors to mimic the natural environment, helping to follow the onset of the disease. Furthermore, UMINHO worked on agent based modelling (ABM) in collaboration with MIRALab. Simulating the molecular events observed in OA helps to reveal the relations in the OA complex system, and may predict the role of the different molecules, forming a novel application in OA research (MIRALab). With ABM, it is possible to establish a more economical, and faster approach to study biological interactions in comparison to laboratory experiments. Another field of work was the development of tissue engineering strategies for meniscus, and biological and biomechanical characterization of the human meniscus and osteochondral tissues as well as the fabrication of optimized tissue engineering scaffolds for meniscus tissue and in vitro and preclinical characterization of the scaffolds.MIRALab has developed tools to handle multi-channel MR image data. They have provided image stitching, muscle tissue labelling and template registration. For musculo-skeletal anatomical modeling, a multi-channel deformable model approach for muscle segmentation was created. Furthermore, the work was focussed on the efficiency by creating a coupled multi-resolution model. Biomechanical analysis of the cartilage mechanical behaviour has been performed and used by Softeco to compute subject-specific cartilage thickness. A subject-specific computational approach to calculate muscle parameters (i.e. pennation angle) was developed and used to personalize the musculoskeletal models (LBB-MHH) and the finite element knee models. A virtual (in silico) model was developed and validated by MIRALab to simulate cellular and molecular reactions during the evolution of osteoarthritis in the cartilage using the biological experiments data provided by UMINHO. The in silico OA model is correlated to the mechanical properties measured by UMINHO and used as input in the FE simulations produced by MIRALab. LBB-MHH has worked on the gait data collection, processing and analyses of the data which has been provided to MIRALab, Welfenlab and CNR-IMATI. The use of generic-scaled musculoskeletal models to detect changes in muscle activations before and after total knee replacement (TKR) has been done as well as the evaluation of simulated muscle activation and internal knee forces of knee enhanced motion activities using generic-scaled versus modified generic-scaled and subject-specific MSK models. An accuracy mathematical calculation was implemented to define a vector of rotation axis of recorded knee extension motion which was used to visually reproduce the kinematic motion and vector of rotation axis in collaboration with Welfenlab.CNR-IMATI has designed and developed several components that form the basis for the development of a CAD system working across multiple scales. These components include: (i) the MSH ontology, with the associated graph-based visualization techniques which make it easier for the users to navigate the knowledge space according to user interests/perspectives; (ii) the MSH Knowledge Base, with the development of several novel methodologies for annotating patient-specific geometries and data with quantitative and qualitative attributes and parameters. Together, the components support comparative analysis of clinical cases and the evaluation of the patient’s follow-up. The components developed by CNR-IMATI contribute to the scientific data visualization component (Welfenlab) by provided a semantics-guided access to data and a profiling of the visualization modalities according to users' interests. The activity on the knowledge framework facilitated the sharing, access to, and integration of the information pertaining to the MSD domain, and provided concrete indications for developing a multi-scale and multi-modal CAD system for storing patient-specific data in order to allow reasoning about musculoskeletal diseases (Softeco).Softeco has addressed the need of measuring the focal femoral cartilage thickness at the weight bearing regions of the knee by developing a reproducible and automatic method from MR images (HUGE) and reconstructed 3D models. Two different ways of evaluating the focal cartilage thickness were implemented. The resulting focal maps originating from the patient-specific simulation data (MIRALab) identified more precise weight-bearing areas and stress region boundaries, and revealed sub-regions that were not considered before. This gives us a considerable advantage when dealing with stress-related OA assessments (UMINHO). A medical data visualization application for comparing segmentation results and measurements was developed: MSH Patient Browser (CNR-IMATI).Welfenlab has worked on the visualization as a graphical representation of data and interaction with the data system employing user friendly methods to support multi-scale exploration. Their Multi-Scale MSD Visualization System allows the presentation of a complete set of multi scalar datasets (LBB-MHH, UMINHO) as individual nodes in a single view while visually indicating semantic connections between the nodes, which is essential in order to understand relations of the contained datasets. The multimodal interaction allows users to feel and manipulate 3D objects in a natural and intuitive manner. By using CNR-IMATI’s ontology, they leveraged the multi-scale biomedical knowledge to present the important aspects of multi-scale data in the Multi-scale MSD Visualization system.The fellows have received training at their hosting institutes and have participated to nine network-wide technical trainings, two workshops and two summer schools. The frequent exchange between partners through secondments lead to fruitful collaboration and the creation of a strong network.The MultiScaleHuman project has been actively disseminating the work of the consortium. Workshops have been organized in the context of conferences like Computer Assisted Radiology and Surgery (CARS, Barcelona, Spain) and Computer Graphics International (CGI, Sydney, Australia). MultiScaleHuman had a booth at the CeBIT exhibition (Hanover, Germany) twice and joined other exhibitions (EXPOSANITA, OCOVA AlpmedNEt Forum, Italy). The presentation of the project at the Festival della Scienza (Genova, Italy) in 2012 and 2014 reached a broad public audience. Two ESR seminars have been organized to present the work to young medical technicians and to encourage woman in science. The research results have been presented at numerous conferences, workshops and meetings. So far, 40 publications have been accepted, with more under review. Two publications have won best paper awards (IEEE BIBM 2014 conference, CBM 2014 workshop at MICCAI). All partners have contributed to a joint book “3D Multiscale Physiological Human”, published by Springer.The impact of the MultiScaleHuman project is motivated by addressing an emerging healthcare challenge of musculoskeletal pathologies that has significant implications in European and worldwide healthcare. The numerous dissemination activities have contributed to increase the awareness of this challenge with the general public and scientific community, while advocating and demonstrating the innovative value of international and multidisciplinary research collaboration as a strategy to answer this challenge. This was reflected by the successful integration of ESRs and ERs in the training network, resulting in an international transfer of knowledge between the different countries involved.
