Artificial intelligence (AI), in its modern form, is profoundly changing many areas, including health. The development of AI for health opens up very promising prospects for improving the quality of care, reducing costs through more personalized care, but also better traceability and improved medical decision-making support. In the field of medical imaging, radiology (in the broad sense) is expected to undergo a significant evolution through the use of machine learning algorithms. The AI4Child project focuses on the development of new medical image analysis methods to assist in the diagnosis and follow-up of patients with cerebral palsy. Cerebral palsy is the most common motor deficiency in children, with a prevalence of 2.1 cases per 1000 births. It affects 17 million people worldwide, and 125,000 in France. This non-progressive disorder causes abnormal patterns of movement and posture, and cognitive and sensory functions can also be impaired. These deficiencies result from structural abnormalities in the brain that occur at different times of eraly brain development. Symptoms are usually apparent before the age of 18 months and the diagnosis is usually confirmed between 13 and 19 months. However, it has been shown that earlier diagnosis would allow for better care of affected children. The objective of AI4Child is therefore to develop and promote new AI-based tools to improve the early diagnosis phase and also to enable monitoring of children's motor development through the emergence of dynamic Magnetic Resonance Imaging (MRI). The core of the project will be based on a methodological development in the continuity of our work in deep learning, and in particular for the understanding of neural network architectures such as residual networks. These new tools will be adapted to the specific issues raised by the study of cerebral palsy, namely: early diagnosis based on cerebral MRI data from the premature baby, high-resolution reconstruction of dynamic in-vivo MRI sequences to measure biomechanical parameters of the child's movements, and finally assistance in rehabilitation through an analysis of walking augmented by in-vivo MRI data. This research work will be carried out at IMT Atlantique and the University Hospital of Brest, within the LaTIM laboratory, in partnership with Philips and the Ildys Foundation. The new knowledge and tools resulting from this work will be disseminated within the scientific communities but also through teaching programmes to the varied public (doctoral students' research training, engineering school, medical school, international summer schools, tutorials). In order to strengthen the links between the various actors, we will continue our action within the European Academy of Childhood Disability (EACD) conference, and more particularly in the organisation of the hackathon bringing together families of disabled children, health professionals, students and researchers. The creation of such synergy in the child-centred ecosystem will promote the impact of the AI4Child initiative at the national and international levels, in research and teaching around AI and its application to health.

Artificial intelligence in Health, and in Ophthalmology in particular, has been very successful over the past four years. The work carried out jointly by LaTIM (Inserm - Research Unit 1101) and Evolucare Technologies, as part of the RetinOpTIC project funded by FUI in 2015, resulted in 2018 in a software program that screens for diabetic retinopathy, a major ocular pathology, based on fundus photographs. The underlying algorithm, whose performance reaches that of a retinal expert, is being deployed in several clinical centers around the world, through the Evolucare Technologies cloud. The success of this solution is partly due to the large amount of data used for learning, namely 760,000 images from 100,000 diabetic patients in the OPHDIAT screening network in Ile-De-France. The common objective of LaTIM and Evolucare Technologies is to extend screening to all pathologies affecting the eye, or visible through the eye (cardiovascular pathologies, neurodegenerative diseases, etc.). In order for each of these conditions to be represented by a sufficient number of examples, the images must come from a much more diverse population than the diabetic population. This is only possible if the data come from several clinical centers. Within the framework of the ADMIRE LabCom, we therefore propose that clinical centers using the Evolucare Technologies cloud participate, if they so wish, in the progressive enrichment of artificial intelligence (AI). In addition to a retrospective analysis of the data from each center, we propose that user feedback on AI diagnoses be used to refine learning over time. In a learning scenario on multi-center health data, it is no longer possible to export all the data to AI researchers. It is also not possible to conduct training in each clinical center, due to limited computing power. We therefore propose to divide learning between clinical centers, on the one hand, and a computer cluster, on the other hand: only abstract information (neural weights, gradients, etc.) will be transfered between the different parties. This will allow each clinical center to keep control over its patients' data and Evolucare Technologies to keep control over the developed AI. However, this raises security issues: in particular, it is necessary to ensure that data flowing between clients and the server, even abstract data, does not allow the extraction of sensitive information about patients or, conversely, about AI models in training. This also raises artificial intelligence challenges. This requires the ability to manage interpretations of varying quality, both from internationally recognized experts of a given pathology and from new AI users; in particular, initial learning should not be forgotten as AI is gradually refined. Finally, it must be possible to manage the variability of clinical centers in terms of collected imaging data (fundus photography, optical coherence tomography, etc.), contextual information and annotations. Solving these various problems will pave the way for the screening of many diseases, and in particular for the screening of rare diseases, for which clinician' experience is necessarily weaker and the contribution of AI is particularly useful. The implementation of this distributed and secure AI platform will also enable LaTIM and Evolucare Technologies to jointly address new clinical problems in the coming years.

Context Each year, about 400 000 patients in Europe and 500 000 in the USA have a Total Hip Athroplasty (THA). 13% of these surgeries lead each year to a hip revision surgery, with an average cost 35% superior to the primary procedure, and with a higher level of morbidity. Multiplying surgical procedures also reduces functional capacities of the patient and significantly increases the risk of infection. One of the main causes of failure is the surgical technique itself, based on statistical criteria defined by retrospective analysis of clinical series. As those criteria present a large variability, they are not always appropriate for a given patient. The goal of arthroplasty procedure is to restore kinematic capabilities close to the original hip joint. However, the planning of such procedure is only based on static criteria. In addition, it is already proven that the kinematic behaviour of the lombopelvic area, specific for each patient, also influences the final stability of the hip implant. Several solutions already exist to guide the surgeon for the placement of hip implants, but those solutions are all based on statistical and static criteria. Solution The goal of this program is to develop a new concept of surgical navigation for total hip arthroplasty that includes a preoperative planning step taking into account dynamic information specifically for each patient. First, a preoperative step measures the pelvic kinematics of the patient and integrates those specific data into the model of a decision support tool. Second, an intraoperative step helps the surgeon to reach the desired position of the implant, thanks to an innovative instrumentation, which secures the surgical procedure and save time. The optimization of the femoral implant is performed with a trial modular neck that can be adjusted thanks to the Smart Screw Driver of Blue Ortho. Objective of this program The goal of this program is to develop an integrated solution comprising (1) an innovative system to measure preoperative dynamic parameters of the patient, (2) an intraoperative guidance software integrating those parameters, (3) an innovative instrumentation and (4) an adjustable implant. A preclinical validation will be performed in order to validate the proof-of-concept. The ratio risk/benefit will be evaluated in a clinical study that is not part of this program in order to determine the delivered medical service. The commercialization will be managed by the industrial partner of this project through an international network already in place.

In recent years, cell-based therapy for cartilage repair has made technological advances to fabricate commercial products and become a promising treatment with artificial joint transplant. Knee cartilage regeneration with mesenchymal stem cells (MSCs) is accomplished through intra-articular injection because of low targeting efficiency of MSCs, the current MSC-based therapy requires a large amount of cells for intra-articular injection or invasive surgery for scaffold implantation above the millimeter size. To improve the lack of targeting efficiency for the present MSC based therapy, this project aims to develop magnetically actuated microrobotic system (EMA) for targeted MSC delivery using multimodal imaging. MagOA will explore the potential of swarms of magnetic microrobots and stem cells for targeted MSC loaded microrobots to stop the degeneration of cartilage during osteoarthrisis (OA) at early stages. Within the project time-frame (4 years), the target is to demonstrate the ground-breaking potential of such a regenerative approach, at a preclinical level. The scientific project objectives will cover the design, modeling, control and clinical development of the microrobotic platform for in-vivo MSCs delivery. A novel robotic EMA system using an eight coil radial structure configuration will be developed compatible visually controlled by clinical medical imaging devices (ultrasonic imaging device and arthroscopy). To make this challenging objective a reality, the project partners will collaborate in the investigation of microrobotics (PRISME, INSA CVL), microrobot functionalization with stem cells (MiNt, INSERM), multimodality navigation (LaTiM, INSERM) and wearable devices (CHU Brest) into a unique workflow. The targeting and regeneration effects of microrobots using the in vivo model of appropriated disease will be fully verified considering the complete clinical environment.

Smart orthopedic implants open up very interesting prospects, particularly for the improvement of post-surgical follow-up. However, nowadays, the technologies available are not adapted to power fully-metallic prostheses used in orthopedics. This project aims to exploit a power transmission solution based on acoustic waves to transmit power in a knee implant. A knee joint model will be developed using new statistical modeling methods integrating acoustic parameters. In addition, the admissible input power levels will be studied to limit the physical mechanisms (thermal, cavitations) and remain below the values set by the standards and used by the commercial ultrasound equipments. This model and the input data will then be used to design, optimize the power transmission solution with both analytical and multi-physics Finite Element modeling methods for a tibial knee implant embedding piezoelectric transducers. We expect the acoustically powered system to receive an amount of electrical power within the 1 mW to 10 mW range at the receiver side while being compliant with medical standards and using commercial ultrasound probes at the transmitter side. Prototypes will be assembled and tested first on five knee phantoms elaborated within the project and then on three cadaveric specimens at the anatomical laboratory of the Brest CHRU. The proofs of concept (PoCs) will then allow to power a new generation of smart orthopaedic implants embedding sensors, that are more robust and more reliable, facilitating industrialization and ultimately allowing better clinical management.