The study of bone structure and its dynamic evolution is of paramount importance in the fields of medicine, biology and biomechanics [1]. Understanding how bones evolve over time at several scales is crucial for diagnosing and treating certain bone diseases, optimising procedures orthopaedic procedures and to advance our knowledge of human anatomy [2]. This research project proposes an original approach integrating 4D imaging, modelling and artificial intelligence to study in depth the multi-scale evolution of bone structure. We will then be able, on the one hand, to elucidate the biophysical mechanisms linking the cellular response to changes in bone tissue and, on the other, to highlight hidden phenomena that may potentially escape physical modelling.

Our main objective is to gain a better understanding of how bones are affected by osteoporosis and drug treatments. The main objectives of this research project are:
O1. Generate a database using 4D imaging techniques. We rely on state-of-the-art four-dimensional (4D) imaging technologies capable of capturing spatial and temporal changes in bone structure, including bone remodelling and other micro-scale alterations, under physiological and pathological conditions. Images of rabbit’s tibia obtained by micro-tomography (µCT) as part of a partnership between QUT and USASK (Canada), will be used to build a longitudinal experimental database (DB1) that is unique in the world. This will be progressively enhanced with other databases to generate an extended database (DBx) incorporating several anatomical sites and animal models. The imaging data generated in this way will provide an ideal working basis for guiding the development and facilitating the calibration and validation of the mathematical models that we intend to build, by offering real references against which the models’ predictions can be tested.

O2. Integration of 4D imaging and image processing. We will develop machine learning (ML) and deep learning (DL) algorithms to analyse bone morphological parameters from µCT data in the DB1 database, in order to assess morphological changes in bone tissue. In addition, we will develop registration algorithms to track bone remodelling units and other structural changes based on the 4D data from DB1. The algorithms will also be evaluated on the extended database (DBx) to assess the model’s ability to adapt correctly to the new data. An essential aspect of this generalisation is to develop methodologies that will allow us to seamlessly integrate data obtained from various imaging devices, animal models and anatomical sites. This will enable us to create coherent and comprehensive spatio-temporal maps of bone remodelling activities, and to compare events at the cellular level with results at the tissue level.

O3. Integration of 4D imaging and multi-scale modelling. We will develop numerical models to predict morphological, structural and mechanical changes in bone tissue in physiological and pathological under physiological and pathological conditions (osteoporosis, drugs, etc.). On the one hand, we will develop small-scale models, such as cellular automata (CA) or agent-based models (ABM), to describe in detail the basic multicellular units (BMU) of bone, also taking into account biochemical and mechanobiological factors. These small-scale models will provide valuable information for developing larger-scale models of bone remodelling. The latter will be used to explore the evolution of the mechanical properties of bone. The small- and large-scale models will be systematically calibrated and validated using data generated in O1 and processed in O2. By linking the models at the two scales, we will be able to build between cellular behaviour at the microscopic scale and structural and mechanical changes in bone at the macroscopic scale. This new approach will enable us to explore the complex interaction between activities at the cellular level and the resulting changes in bone architecture.

The research products of this project are fundamental in nature, but will eventually be used in clinical practice. They are divided into three parts:
1/ generation of a unique database of 4D images;
2/ the development of ML/DL algorithms dedicated to 4D image processing;
3/ advancing knowledge of bone remodelling. A conference will be organised in Australia at the end of 2025 to disseminate and promote these results.
This project will also strengthen the collaboration between MSME/UPEC and QUT, as well as structuring a group of young researchers around the modelling of the functional adaptation of bone tissue. What’s more, the expertise created in this project could eventually be extended to other biological tissues.
Finally, this project will serve as a stepping stone for the creation of a “Life Sciences Engineering” Scientific Community within AFRAN, led by V. Sansalone (who is on the board of AFRAN-France) and P. Pivonka.

UMR8208 – Modélisation et simulation multi-échelle (MSME) 

Australia