ASP projects

GAP Image-Guided experimental and computational Analysis of fractured Patients

The study of bone damage processes at different scales is essential for understanding fracture mechanisms, which are mostly induced by a trauma or a pathology (such as osteoporosis). Early diagnosis is critical for reducing the burden of bone fractures on the health care system and the economy. This problem is deeply rooted in our society and it will grow more and more because of the increase in average age; in fact, according to data from the Italian Ministry of Health, 40% of the total Italian population, mostly after the age of 65, will have a fracture of the femur, vertebrae or wrist.
However, among the research and experiments on the bone mechanism, the role of bone micro-scale features is yet unclear and the opinions about the presence of small cavities called lacunae are contrasting: lacunae, in the first instance, are stress concentrators that contribute to strength reduction; bone is a damage-tolerant material and lacunae contribute to toughness by deviating crack propagation.
Shedding light on this still fuzzy scenario could be the key to developing early diagnosis methods, avoiding and preventing large numbers of fractures and easing the burden on the health care system. In order to reach our goal, a deeper comprehension of damage initiation and propagation at micro-scale was needed and our approach to achieve it was studying the phenomenon through both experimental tests and computational analysis.
The project has been structured in the following phases:
● Problem definition and Review of the state of the art: A deep study of bone structure has been the starting point. Later, we have moved to problem definition by investigating the economic and psychosocial burden of osteoporosis.
● Preliminary activities for experimental testing: First of all, human femoral heads samples have been collected prior authorization from the ethics committee of Gruppo San Donato Foundation and approval of the patients. At the same time, we have designed and realised a micro-compression device for dynamic image-guided failure assessment of human trabecular bone microstructure and bone micro damage according to our requirements.
● Experimental testing: Compression tests under displacement control have been performed on human femur head samples. The tests have been performed at the ELETTRA Synchrotronpremises in Trieste where a technology able to couple μ-CT to synchrotron sources can be found.
● Image post-processing: We designed an AI tool capable of recognizing cracks and lacunae automatically through the use of a Convolutional Neural Network. This AI tool has solved our problem since it is suitable to deal with classification and detection issues.We have realized, trained, and tested two algorithms, one for cracks detection and the other for lacunae recognition.
● Computational analysis: Owing to the collaboration with ETH, we have succeeded in implementing a sophisticated computational damage model, making more understandable cracks’ history, initiation and propagation and clarifying lacunae’s role in these processes.
● Bio-inspired structures: The ability of bone to withstand fracture under a wide range of stresses inspires structural uses, while the high surface area to volume ratio and pore connectivity of bone architecture present fascinating aesthetic options. Thus, we have decided to write an article that explores the current state of the art in bone-inspired applications in product design, architecture and fashion, discussing further technological developments (“Down to the Bone: A Novel Bio-Inspired Design Concept” on Journal MDPI).
● Enhancement of convolutional neural network tool: Our AI tool has met the approval of clinicians and seems to have potential for further developments, not only in analysing micro-scale bone images, but could also adapt in many different scenarios. For this reason, we have decided to write a scientific article on it, that is currently in the process of submission.

The main result of our project is a better comprehension of micro-cracks initiation and propagation and of the role played by the lacunae. Indeed, both experimental tests and computational analysis have demonstrated that lacunae act as crack attractors and deviate the crack path. Another strong point of our project is the quality and the quantity of bone sample images at micro-scale that represent a great resource for future studies. Moreover, the convolutional neural networks developed have strongly fastened the image analysis and they are extremely innovative and very promising tools. The multiscale framework of the bone has proved to be very elaborate and resourceful, not only from a biomedical point of view, but also architects, product and fashion designers have appreciated the properties of this structure as the article we have published demonstrates.

Principal Academic Tutor:
Laura Maria Vergani (Politecnico di Milano)

Academic Tutor
Federica Buccino (Politecnico di Milano)