Fluctuations microscopy for functional imaging of organoids
Phase contrast microscopy and fluorescence microscopy are the two pillars of modern biological imaging. Phase contrast reveals the morphology of the sample, while fluorescent labeling provides specificity to the process of interest. In both cases, the image is the average value of the measured signal. In this thesis, we propose to focus not on the average value, but on the fluctuations observed in phase contrast. This new contrast will be called Fluctuations Imaging. The fluctuations arise from the active and passive transport phenomena that characterize cellular machinery, and it can be assumed that the level of fluctuations is correlated with cellular activity. The objective of the thesis is to detect phase contrast fluctuations, quantify them, and link them to a process of interest using machine learning methods. The object of study will be lymphocyte activation, which is a critical parameter for monitoring rejection in certain patients with type 1 diabetes who have undergone islet transplantation. Fluctuations Imaging would enable tracking without labeling, simplifying the monitoring protocol. The expected work is (i) optimizing a phase contrast microscope to detect fluctuations, (ii) analyzing image sequences to quantify them, and (iii) implementing the developed method on various biological models, some of which will be pancreas-on-a-chip organs. This thesis, at the intersection of instrumentation, biophysics, and biology, is intended for a student with a background in optics, physics, or equivalent, with a good knowledge of image processing and a strong interest in applications in biology and health.
Development of injectable adhesive hydrogels for the treatment of retinal tears
Retinal tears then detachment, a serious eye condition (20–25 cases per 100,000 in France each year), requires urgent surgery. Current treatments involve removing the vitreous, using gas as a tamponade agent, and sealing tears with laser. However, this method presents drawbacks, including patient restrictions (e.g., prolonged lying down) and complications (e.g., cataracts). Injectable hydrogels are being explored as alternative tamponade agents, but they do not display adhesive properties to suture the tears and laser treatment is still required. Surgical glues have also been tested, but cyanoacrylate-based adhesives are toxic, fibrin-based sealants are hard to use in the eye, and current hyaluronan (HA)-based materials lack sufficient stability and adhesion.
This PhD project aims to develop a sterile, injectable HA-based hydrogel with strong adhesive properties to seal retinal tears. Key requirements include biocompatibility, injectability (30G needle), tissue adhesiveness (1.5–3.7 N), and rapid delivery (within 1 hour). Our group has previously developed an injectable HA hydrogel with dynamic crosslinking, offering long-term stability, biocompatibility, and optical transparency. To confer it with tissue-adhesion properties, two strategies will be tested: (1) addition of tissue-adhesive tannic acid in the hydrogel formulation, or (2) grafting tissue-adhesive groups onto the HA backbone. The hydrogel will be tested for its biocompatibility and adhesiveness in preclinical eye models.
This innovative hydrogel could simplify retinal surgery, reduce complications, lower costs, and improve recovery. Beyond retinal repair, it may have applications in cornea surgery and other medical fields.
Implementation of TFHE on RISC-V based embedded systems
Fully Homomorphic Encryption (FHE) is a technology that allows computations to be performed directly on encrypted data, meaning that we can process information without ever knowing its actual content. For example, it could enable online searches where the server never sees what you are looking for, or AI inference tasks on private data that remain fully confidential. Despite its potential, current FHE implementations remain computationally intensive and require substantial processing power, typically relying on high-end CPUs or GPUs with significant energy consumption. In particular, the bootstrapping operation represents a major performance bottleneck that prevents large-scale adoption. Existing CPU-based FHE implementations can take over 20 seconds on standard x86 architectures, while custom ASIC solutions, although faster, are prohibitively expensive, often exceeding 150 mm² in silicon area. This PhD project aims to accelerate the TFHE scheme, a more lightweight and efficient variant of FHE. The objective is to design and prototype innovative implementations of TFHE on RISC-V–based systems, targeting a significant reduction in bootstrapping latency. The research will explore synergies between hardware acceleration techniques developed for post-quantum cryptography and those applicable to TFHE, as well as tightly coupled acceleration approaches between RISC-V cores and dedicated accelerators. Finally, the project will investigate the potential for integrating a fully homomorphic computation domain directly within the processor’s instruction set architecture (ISA).
Learning to focus: Physics-Informed Deep Learning for Super-Resolved Ultrasonic Phased-Array Imaging
This PhD aims to develop a new class of ultrasonic focusing methods for phased-array imaging by combining deep learning, physics-based modeling, and optimal transport theory. The first research axis introduces a reweighted, probabilistic extension of the Total Focusing Method (TFM), where per-isochrone focusing weights are iteratively estimated by a shared convolutional network and normalized using a neural time-of-flight field. This iterative, differentiable framework enables more adaptive, interpretable, and robust imaging in heterogeneous or uncertain media.
The second axis proposes a full reformulation of TFM as a Wasserstein barycenter problem, in which each partial image is modeled as an empirical distribution in a joint space of spatial coordinates and ultrasonic amplitude. A physically meaningful transport cost, based on geodesic distances that minimize time-of-flight variations with respect to selected emitters, encodes the acoustic geometry directly in the metric. The resulting grid-free barycenters yield sharp, physically consistent reflector localization and open new opportunities at the interface between optimal transport and ultrasonic phased-array imaging. Overall, the thesis aims to merge physics, machine learning, and geometric optimal transport to formulate next-generation reconstruction methods for ultrasonic imaging.
Advancing Health Data Exploitation through Secure Collaborative Learning
Recently, deep learning has been successfully applied in numerous domains and is increasingly being integrated into healthcare and clinical research. The ability to combine diverse data sources such as genomics and imaging enhances medical decision-making. Access to large and heterogeneous datasets is essential for improving model quality and predictive accuracy. Federated learning is currently developed to support this requirement offering an alternative by enabling decentralized model training while ensuring that raw data remains stored locally at the client side. Several open-source frameworks integrate secure computation protocols for federated learning but remains limited in its applicability to healthcare and raises issues related to data sovereignty. In this context, a French framework is currently developed by the CEA-LIST, introduces an edge-to-cloud federated learning architecture that incorporates end-to-end encryption, including fully homomorphic encryption (FHE) and resilience against adversarial threats. Through this framework, this project aims to deliver modular and secure federated learning components that foster further innovation in healthcare AI.
This project will focus on three core themes:
1) Deployment, monitoring and optimization of deep learning models within federated and decentralized learning solutions.
2) Integrating large models in collaborative learning.
3) Developing aggregation methods for non-IID situation.
Optical intradermal sensing via instrumented microneedles
Cortisol plays a central role in regulating the circadian cycle and in many essential physiological processes such as energy metabolism and immune response. Conventional monitoring of cortisol relies on single blood or saliva samples, which do not accurately reflect the temporal dynamics of its secretion. It is therefore necessary to develop innovative approaches that enable continuous, minimally invasive, and reliable measurement of cortisol concentration in patients.
The doctoral project aims to develop an original optical instrumentation system coupled with microneedles functionalized with fluorescent aptamers for continuous, minimally invasive intradermal monitoring of cortisol without the need for sampling. Within this framework, the PhD candidate will be responsible for designing and sizing the future optical microneedles intended for cortisol detection. They will set up the experimental systems required to characterize the optical microneedles fabricated within the department and test their performance in a representative environment. Finally, the PhD candidate will develop a comprehensive data processing and analysis methodology to identify the key parameters that establish a quantitative relationship between the collected signals and cortisol concentration. Altogether, this work will contribute to the development of an innovative measurement device based on cutting-edge optical emission and detection technologies available at CEA Leti, combining precision, sensitivity, compactness, and thus compatibility with in situ use.
Development of 4D-STEM with variable tilts
The development of 4D-STEM (Scanning Transmission Electron Microscopy) has profoundly transformed transmission electron microscopy (TEM) by enabling the simultaneous recording of spatial (2D) and diffraction (2D) information at each probe position. These so-called “4D” datasets make it possible to extract a wide variety of virtual contrasts (bright-field imaging, annular dark-field imaging, ptychography, strain and orientation mapping) with nanometer-scale spatial resolution.
In this context, 4D-STEM with variable beam tilts (4D-STEMiv) is an emerging approach that involves sequentially acquiring electron diffraction patterns for different incident beam tilts. Conceptually similar to precession electron diffraction (PED), this method offers greater flexibility and opens new possibilities: improved signal-to-noise ratio, faster two-dimensional imaging at higher spatial resolution, access to three-dimensional information (orientation, strain, phase), and optimized coupling with spectroscopic analyses (EELS, EDX).
The development of 4D-STEMiv thus represents a major methodological challenge for the structural and chemical characterization of advanced materials, particularly in the fields of nanostructures, two-dimensional materials, and ferroelectric systems.
Infertility is a growing problem in all developed countries. The standard methods for the diagnostic of male infertility examine the concentration, motility and morphological anomalies of individual sperm cells. However, 40% of male infertility cases remain unexplained with the standard diagnostic tools.
In this thesis, we will explore the possibility to determine the male infertility causes from the detailed analysis of 3D trajectories and morphology of sperms swimming freely in the environment mimicking the conditions in the female reproductive tract. For this challenging task, we will develop a dedicated microscope based on holography for fast imaging and tracking of individual sperm cells. Along with classical numerical methods, we will use up-to date artificial intelligence algorithms for improving the imaging quality as well as for analysis of multi-dimensional data.
Throughout the project we will closely collaborate with medical research institute (CHU/IAB) specialized in Assisted Reproductive Technologies (ART). We will be examining real patient samples in order to develop a new tool for male infertility diagnosis.
Development and multiparametric monitoring of a microfluidic chip of the blood-brain barrier model
The blood-brain barrier (BBB) protects the brain by controlling exchanges between the blood and nervous tissue. However, current models struggle to accurately reproduce its complexity. This thesis aims at developing and evaluating a microfluidic chip of BBB model incorporating a real-time monitoring system that combines simultaneous optical and electrical measurements. The device will enable the study of permeability, transendothelial resistance and cellular response to various pharmacological or toxic stimuli. By combining microtechnologies, cell co-cultures and integrated sensors, this model of biological avatar will offer a more physiological and dynamic approach than conventional in vitro systems to improve understanding of the diffusion/permeation phenomena of therapeutic molecules. This project will contribute to the development of predictive tools for neuropharmacology, toxicology and research into neurodegenerative diseases.
Integrated material–process–device co-optimization for the design of high-performance RF transistors on advanced nanometer technologies
This PhD research focuses on the integrated co-optimization of materials, fabrication processes and device architectures to enable high-performance RF transistors on advanced nanometer-scale technologies. The work aims to understand and improve key RF figures of merit—such as transit frequency, maximum oscillation frequency, noise behaviour and linearity—by establishing clear links between material choices, process innovations and transistor design.
The project combines experimental development, structural and electrical characterization, and advanced TCAD simulations to analyse the strengths and limitations of different integration schemes, including FD-SOI and emerging 3D architectures such as GAA and CFET. Particular attention will be given to the engineering of optimized spacers, gate stacks, junction placement and epitaxial source/drain materials in order to minimize parasitic effects and enhance RF efficiency.
By comparing planar and 3D device platforms within a unified modelling and characterization framework, the thesis aims to provide technology guidelines for future generations of energy-efficient RF transistors targeting applications in 5G/6G communications, automotive radar and low-power IoT systems.