The AI4CODE project brings together 6 research team with strong expertise in the design, decoding and standardization of forward-error-correction codes. The aim is to develop skills in artificial intelligence and machine learning, and to explore how learning techniques can contribute to the improvement of code design methods (by using less parameters, more relevant heuristics, producing stronger codes) and decoders (better performance, reduced complexity or energy consumption), on selected scenarios of practical interest for which a full theoretical understanding is still lacking. The proposed methodology is to augment legacy design methods and decoders with learning capabilities or decision support systems wherever relevant, rather than replacing them by a generic, black-box neural network, so that we can inspect the trained solutions and try to infer why they work better. Our ultimate goal is to obtain new theoretical hindsight that could translate into better codes and decoders.

Ventricular arrhythmias are a major cause of sudden cardiac death in Europe (350,000 deaths / year). The majority of these deaths cannot be anticipated and prevented due to the low sensitivity of current risk criteria. Cardiac mapping of surviving patients has shown the presence of distinctive electrical signals in areas generating arrhythmias. These signals are currently recorded by invasive measurement (catheters) as they are not perceived by electrocardiography or electrical mapping on the surface of the thorax. Detection of the magnetic components of these signals would allow a vector measurement including currents electrically invisible consequence of an altered myocardial zone. The aim of the MAESTRO project is to implement a network of ultrasensitive magnetometers on models of pathological hearts (ex-vivo and in-vivo) in order to develop a non-invasive method of identifying signals associated with a high risk of sudden cardiac death.

Diabetes is a major public health problem in the world. In 2018, this pathology is still responsible of severe complications (end-stage renal failure, dialysis, blindness). The treatment for type 1 diabetic (T1D) patients and for approximately 15% of insulin-treated type 2 diabetic patients relies partly on: the daily administration of insulin, the multi-daily control of their blood glucose, their dietary carbohydrates, and the daily adjustment of their insulin therapy. However, the treatment for diabetes remains insufficient to restore optimal metabolic control, and patients continue to develop complications linked to their diabetes. In addition, some T1D patients develop a particular form of diabetes, unstable diabetes that exposes them to: major changes in their blood glucose levels and the occurrence of severe metabolic events. The future of the treatment of diabetes, especially type 1, is turned towards cell therapy or islet transplantation of Langerhans, which proposes to transplant to the T1D patient suffering from unstable diabetes of the islets of Langerhans, functional anatomical units containing insulin secreting cells. This technique has been shown to be efficient in restoring optimal glycemic control in humans with currently over 1000 islet transplanted patients. Insulin weaning can be achieved in half the patients after 1 year. Islet transplantation remains limited to a limited number of patients because of the sometimes severe risks of immunosuppressive therapy, which remains essential. There is also a loss of long-term graft functionality and a shortage of islet graft that limits the applicability of the technique to a larger number of patients. The future of the islet transplantation technique will have to be without immunosuppressive therapy, finding an alternative source to human islets, and enabling more sustainable graft functionality. The BIOCAPAN (BIOactive implantable CApsule for PANcreatic islets immunosuppression free therapy) consortium has been developing since 2015 a translational research project funded by European H2020. This project aims at the industrial development on a large scale, of microcapsules consisting of islets encapsulated by a microfluidic technique in the presence of different biopolymers offering the prospect of being able to graft these islets without any immunosuppressors while also increasing their functionality. The current BIOCAPAN project is, to our knowledge, the first project allowing the GMP and industrial production of encapsulated islets and combining all the biomaterials composing the biocapsule. During the BIOCAPAN project, the availability of human islets, the heart of the capsule composition, has emerged as a major limiting factor, leading to having recourse to an alternative source to the islets. The CAPITAL project is based on the innovations developed during the BIOCAPAN project and proposes to associate an alternative source of cells to the human islets by proposing the use of insulin-secreting cells obtained from induced pluripotent stem cells, the use of islets genetically modified pork or the use of biopolymers capable of releasing insulin. Thus, the translational research project CAPITAL, consisting of scientists, industrials and clinical research team expert in T1D cell therapy and aiming at the development and transplantation of insulin-secreting biocapsules in diabetic patients (80 million worldwide) is fully in line with the call for projects H2020 sc1-bhc-07-2019.

This project aims to develop a brand new light sensor array for biomedical applications. It is designed to meet the criterion of "non-invasive blood glucose monitoring". It is thought to meet the requirements of the treatment of diabetes type I. In addition it is built around a scalable technology: application to blood glucose is one of possible applications among others (certainly highly bankable and highly wished by patients nowadays). It will be fully reprogrammable, resizable and geometrically reformattable. It is structured as a real non-invasive imaging and dosing device for organic compounds in living tissues. So it can be adapted to the assay of biological molecules involved in other diseases. The main of its features is to be wireless which enables real-time reporting toward the physician (in charge of the patient) of any drift in monitored parameters. It targets the mobile medicine and digital health (mHealth and E-Health). Biometric data such as heart rate, blood saturation, respiration, movement and many other physiological signals will also be easily analyzed and transmitted. These data are currently not monitored continuously and this application will improve the well-being of patients. Indeed it will reconsider the scope of health and medical care by bringing closer outpatients to healthcare professionals (hospital "on demand"). A sticky patch as we plan to do is the only sensor shape that can provide continuous and reliable analysis of a parameter (this trend has been validated by the mayo clinic in diabetes). The challenge will be to find the right technology to meet the requirement of medical standards level. The conformation of this sensor is the main concern for whom wants to get a good measurement accuracy because of the skin properties (inhomogeneous medium with light absorbance and light scattering). Unlike other techniques currently used, we will rely on the BEER-LAMBERT law in all wavelengths of the visible and Infrared spectrum while designing a sensor with multiple measurement points. Comparatively, the existing technique is based on two measuring points and up to four wavelengths. In order to decrease calculations when processing the signal and to "select" a fixed depth of the skin where stands the vessels, we will use the polarization properties of light among others. Coupled with this, spectroscopy is very promising for clinical applications because it is portable, relatively inexpensive, and the light is non-ionizing. near-infrared spectroscopy (NIRS) will be very useful because in this wavelength range, each biochemical compound has a unique "fingerprint". This is already used outside the medicine for detection and quantitation purpose (biochemistry, fermentation, pharmacy, ..). NIR spectrum penetrates deep into the skin because it is very poorly absorbed by the latter compared to other wavelengths. CONCLUSION: Our goal in this project is to develop a new imager that integrates both the illumination source and the detector (CMOS). The device should hold the surface of a thumb. However, the limited efficiency of absorption of silicon involves the need to develop a new sensor. A new architecture is being considered. To obtain a high S/N ratio for the whole visible spectrum, while retaining a high spatial resolution. Besides its metabolic capabilities, its architecture will take real pictures of the tissue architecture. It will open the door to new analytical techniques. All this should lead to the development of an imaging array for medical applications with high spatial & chromatic resolution. This is the purpose of the ANR.

High speed imaging is a booming activity with the ideal application of CMOS technology imagers. It makes it possible to acquire a fast single event at a fast sampling and frame rate and to observe it at a reduced frame rate. It finds many applications in motion analysis, explosives, ballistic, biomechanics research, crash test, airbag deployment, manufacturing, production line monitoring, deformation, droplet formation, fluid dynamics, particle, spray, shock & vibration, etc. High speed video imaging is currently driven by some industrial manufacturers such as Photron, Redlake, Drs Hadland, which design their own sensors. The current industrial most efficient imagers offer a speed of 22,000 frames per second (fps) for a spatial resolution of 1280x800 pixels, i.e. 22 Gpixel/s. This speed is not restricted by the electronics of the pixel but by the sensors chip inputs/outputs interconnections. The conventional operation mode based on extracting the sensor data at each acquisition of a new image is a real technological barrier that limits the scope of high speed cameras to the study of transient phenomena that last for a few hundred microseconds. The FALCON project's main goal is to overcome this technological barrier, increasing the acquisition speed by three orders of magnitude by proposing a sensor capable of taking up to 100 million fps while increasing the sampling rate up to 10 TeraPixel/s. To accomplish this, the classical approach of extracting image sensor should be abandoned in favor of a new one which makes it possible to eradicate the inputs/outputs bottleneck. Several studies mention the realization of high-speed image sensors based on the principle of "burst" imagers (BIS Burst Image Sensor). Since it is impossible to get the frames out of the sensor as they are acquired, the idea is to store all the images in the sensor and execute the readout afterward, after the end of the event to be recorded. So far, all the developed BIS based on this principle use a totally analog approach in the form of a monolithic sensor. The size of the embedded memory is generally limited to a hundred frames, the pixel pitch is around 50 µm and the acquisition rate is in the order of 10 Mfps for large 2D arrays. Furthermore, research works mention little data about the signal to noise ratio (SNR), but the leakage current of the storage capacities degrades the signal quality and the effect is more noticeable when the readout duration is high, i.e. when the number of stored images is large. This phenomenon limits, once again, the number of storable images in analog BIS forms. In general, a maximal SNR of 45 dB is obtained. The FALCON project is based on a device concept in total disruption with previous works, by implementing the possibilities offered by the emergent microelectronics 3D technologies in order to increase the performance of this type of sensor while also adding more features to it. A PhD work started in 2012 in collaboration between the CEA Leti and the ICube laboratory helped to determine an optimal sensor architecture that takes advantage of the 3D technology. A particularity of the proposed architecture is the in-line analog to digital conversion at full speed. This study shows that the proposed new approach increases the number of stored images, while increasing the signal to noise ratio. It has also brought light to the potential problems of heat dissipation inherent to both fast circuits and 3D technologies. The methodological aspects of the design are also at the center of the project seeing that architecture/partitioning and electronic/thermal co-designs are necessary to carry out this type of conception. New tools and methods for the design of integrated heterogeneous systems are needed. The ultimate objective of the project is a high definition 1200x1200 pixels, 10 Mega fps with more than 1000 frames embedded digital memory. The project is pushing the performances of all the system bricks to the state of the art.