Biomedical Master Thesis

Promotor, co-promotor, advisor : amin.shavandi@ulb.be, - , man.li@ulb.be

Research Unit : 3BIO-BIOMATTER

Description

attached pdf document

Promotor, co-promotor, advisor : amin.shavandi@ulb.be, - ,

Research Unit : 3BIO-BIOMATTER

Description

attached pdf document

Promotor, co-promotor, advisor : amin.shavandi@ulb.be, - , amir.abrishami@ulb.be

Research Unit : 3BIO-BIOMATTER

Description

attached pdf document

Promotor, co-promotor, advisor : amin.shavandi@ulb.be, - , Man.Li@ulb.be

Research Unit : 3BIO-BIOMATTER

Description

attached pdf document

Promotor, co-promotor, advisor : ilias.el.makrini@vub.be, - , Tom Turcksin

Research Unit : BRUBOTICS R&MM

Description

Object Detection for Force Estimation in Biomechanical Assessment

The project aims to estimate interaction forces during human-object interaction using vision-based object detection and tracking, improving biomechanical assessments and ergonomic analysis.

Context

The project is done in collaboration with AugmentX (www.augmentx.be). It uses a pair of RGB-D (depth) cameras to capture hand-object interactions during physical tasks.

Objective

To develop a computer vision system that detects and tracks objects in real time and estimates the forces applied by the user’s hands based on object properties such as shape, mass, and acceleration.

Methods

Different methods are to be tested, namely:

*Literature review on object detection, tracking, and force estimation techniques;

*Implementation of a detection and tracking system using RGB-D cameras;

*Development of an algorithm to estimate interaction forces from object motion and known properties;

*(Optional) Extending the system to work with moving cameras;

*Validation against reference force data in controlled experiments.

Prerequisite

• Matlab,
• Python or C++

Contact person

For more information please contact: tom.turcksin@vub.be

Promotor, co-promotor, advisor : ilias.el.makrini@vub.be, - , Ilias El Makrini

Research Unit : BRUBOTICS R&MM

Description

Development of a Modular Simulation for Ergonomic Task Analysis in Unity3D

The project aims to build a proof-of-concept simulation and post-processing pipeline for ergonomic analysis using Unity3D for human motion generation and MATLAB for data processing and visualization.

Context

The aim is to enable rapid prototyping and evaluation of human task performance based on simulated movements, especially in industrial and assembly contexts.

All work will be conducted in simulation (Unity3D), so no lab hardware is needed. Motion data will be generated from known task constraints and processed offline in MATLAB. The developed architecture will serve as a basis for more advanced biomechanics or real-time optimization studies in the future.

Objective

To develop a modular software that simulates human task execution in Unity3D and extracts joint-level data for ergonomic post-processing in MATLAB. The simulation will be used to analyze joint angles and posture scores, forming the foundation for future real-time ergonomic feedback systems.

Methods

The project will consist of several key phases:

• Architecture design: A software architecture will be designed to integrate Unity3D (for simulation) with MATLAB (for analysis). This includes task definition, movement simulation, and data export structures.

• Avatar & task setup in Unity: A human avatar will be imported into Unity3D (e.g., from Mixamo), and a few sample industrial tasks (e.g., object reach, lifting) will be animated using inverse kinematics or keyframe animations.

• Joint data extraction: Scripts will be implemented in Unity to log joint angles and hand trajectories during task execution, exported in a format readable by MATLAB.

• Post-processing in MATLAB: MATLAB scripts will be developed to calculate simple ergonomic scores (e.g., RULA, EAWS approximation) and plot joint usage statistics.

• GUI prototype: A basic interface in Unity or MATLAB that allows users to select which metrics or body segments to visualize.

This project can be carried out by two students working in parallel (e.g., one focusing on Unity simulation, the other on MATLAB post-processing).

Prerequisite

• C#, C++,
• Matlab

Contact person

For more information please contact : ilias.el.makrini@vub.be

Promotor, co-promotor, advisor : fabrizio.pucci@ulb.be, Prof. Marianne Rooman,

Research Unit : COMPUTATIONAL BIOLOGY AND BIOINFORMATICS

Description

The prediction of drug response in cancer cell lines has become a key strategy for identifying molecular factors that influence treatment efficacy. Datasets such as Genomics of Drug Sensitivity in Cancer (GDSC) have provided fundamental information to link genomic and transcriptomic features to drug sensitivity. In this master's thesis, we aim to develop a machine learning framework to predict drug responsiveness by integrating multiple layers of biological information. Specifically, we will extract and engineer features from genomic and transcriptomic data, while also incorporating protein structural information to enhance the biological interpretability of the model. Beyond modeling drug response in cancer cell lines, we will extend our approach to a more clinically relevant scenario by accounting for tumor clonal heterogeneity—a key challenge in personalized oncology. In this phase, we will identify driver mutations from patient-derived genomic cohorts and assess their potential impact on drug sensitivity. This will be achieved by predicting how these mutations alter protein function and, in turn, influence treatment response. The ultimate goal of this work is to move toward a precision oncology framework, where we can computationally predict the most effective drugs or drug combinations for individual patients based on their unique molecular profiles.

For inquiries, please contact: Fabrizio.Pucci@ulb.be

Promotor, co-promotor, advisor : fabrizio.pucci@ulb.be, Prof. Marianne Rooman,

Research Unit : COMPUTATIONAL BIOLOGY AND BIOINFORMATICS

Description

Maintaining global health requires the development of generic and versatile technologies that allow fast and effective responses to the large variety of disorders, in particular cancer and emerging infectious diseases. Among these, monoclonal antibodies (mAbs) play an important role. Indeed, antibodies can bind antigens, such as bacterial or viral proteins or proteins expressed in cancer cells, and trigger the human immune response. In this project, we will build a bioinformatics pipeline for the design of mAbs that can bind with high affinity a specific target antigen. For that purpose, we will rely on experimentally characterized antibody-antigen complexes and their binding affinity values, detect informative sequence- and structure-based features, and combine them into a predictor using artificial intelligence techniques. The mAbs that we will design will be tested in vitro by experimental collaborators. While the project is focused on the design of a generic pipeline, it can be applied to specific case studies such as chronic lymphocytic leukemia, on which we are currently collaborating with cancer immunologists (Institute Bordet, BE), or infectious in collaboration with the CER research center.

Promotor, co-promotor, advisor : gilles.bruylants@ulb.be, Maurice Retout,

Research Unit : EMNS

Description

Photoacoustic Imaging (PAI), the fastest growing biomedical imaging modality in the last decade, has the potential to significantly impact the field of nanomedicine. It is non-ionizing, non-invasive and uses a nanosecond pulsed laser to generate pressure waves that can be detected by conventional ultrasound transducers. Because PAI uses a light-in-sound-out approach, it has the strengths of ultrasound, i.e. good tissue penetration, real-time monitoring, low cost and high spatial resolution, but also the high contrast, specificity and sensitivity of optical methods. Although endogenous contrast agents such as oxygenated or deoxygenated hemoglobin and melanin can be used, PAI still lacks exogenous contrast agents, which could increase sensitivity and allow targeting of specific cells (such as cancer cells). This improved diagnostic capacity could also be combined with therapeutic activity, creating new and promising theragnostic platforms. The EMNS laboratory is involved in the development of such functionalized nanomaterials based on gold nanorods and copper sulfide nanoparticles.

Students will synthesize and characterize the functionalized nanostructures (UV-Vis, IR, DLS, TEM, …) and evaluate their potential as contrast agents for PAI and photothermal transducers.

Contact person For more information please contact : Gilles.Bruylants@ulb.be; Maurice.Retout@ulb.be

Promotor, co-promotor, advisor : serge.schiffmann@ulb.be, Olivier Debeir, Christophe Varin

Research Unit : LAB. NEUROPHYSIOLOGY

Description

Title Analysis of large-scale calcium imaging recordings of neuronal activity in relation with naturalistic and acquired behaviors.

Context The basal ganglia play a key role in the control of both goal-directed behaviors and natural, self-paced behaviors. The proper initiation and execution of these behaviors rely heavily on appropriate functioning within the basal ganglia. Indeed, basal ganglia dysfunction is at the core of various disorders, including Parkinson’s disease, autism spectrum disorders, and schizophrenia (Gunaydin and Kreitzer, Annu. Rev. Physiol., 2016). The striatum, which is the main entry nucleus of the basal ganglia, consists of two types of striatal projection neurons (SPNs) that differ based on their expression of either dopamine D1 or D2 receptors and their respective direct or indirect projections to the output nuclei of the basal ganglia (dSPNs or iSPNs). Previous evidence has led to divergent conclusions on the respective engagement of both pathways during the execution of spontaneous actions as well as during the acquisition of new goal-directed instrumental behaviors. Our team recently proposed an updated model for motor encoding among SPNs in the dorsal striatum that relies on the congruent activation of dSPNs, which encode multiple accessible behaviors in a given context to promote these behaviors, and iSPNs, which encode for and inhibit competing behaviors (Varin et al., Nat. Comm., 2023). As a result, the coactivation of specific subsets of behavior-promoting dSPNs and behavior-suppressing iSPNs alongside specific inhibition of subsets of iSPNs allowing behavior expression would result in the selection and execution of only one motor program.

Project This work aims at validating and expanding our understanding of the organization of neuronal activity among dSPNs and iSPNs during naturalistic and learned goal-directed opreant behaviors using in vivo microendoscopic recordings of neuronal activity recorded through calcium indicators expressed specifically in dSPNs and iSPNs. The project will rely on the analysis of already acquired datasets obtained in mice learning an operant goal-directed task and on the acquisition and analysis of new recordings harvesting in mice submitted to a set of behavioural experiments (e.g. self-paced exploration of open fields of different shapes, sizes, proximal cues, elevated plus maze, light-dark room). The goal of the project will be to decipher and compare the encoding of actions between dSPNs and iSPNs and how their respective encoding properties for a given action evolve during learning and when external contingencies are modified.

Promotor, co-promotor, advisor : olivier.debeir@ulb.be, jennifer.dhont@hubruxelles.be, younes.jourani@hubruxelles.be

Research Unit : LISA - IMAGE

Description

Project title

The project aims to solve an open issue in a certain domain of application.

Background

Tumor segmentation in digital pathology plays a crucial role in breast cancer diagnosis and prognosis [1], [2]. Precise delineation of malignant epithelial regions in hematoxylin and eosin (H&E)-stained or immunohistochemistry (IHC)-stained slides enables downstream analyses, such as cellularity estimation and biomarker quantification for diagnostic pathological examination, therapeutic response assessment, treatment selection, and survival prediction [3]–[8]. Deep learning-based segmentation approaches overcome the inefficiency of manual assessment, enabling high-throughput analysis of histopathological datasets. However, current approaches predominantly rely on supervised learning, which requires labor-intensive pixel-level manual annotations that are impractical at scale [9]–[11].

Weakly supervised learning has emerged as a promising alternative, leveraging coarse-grained labels to reduce annotation burdens. Yet, existing solutions are constrained by the restriction to whole-slide image (WSI)-level classification [12], reliance on partial cell-level annotations [13], and unproven generalizability across diverse breast cancer cohorts and staining protocols [14], [15]. These challenges underscore the need for a weakly supervised segmentation method that is trained using only image-level annotations while achieving pixel-level precision in malignant epithelium delineation and generalizing to heterogeneous breast cancer datasets.

Specific tasks

• Literature study to get familiar with the different topics.

• Perform data preprocessing, including extracting patches from whole slide images, applying color deconvolution to separate the Hematoxylin stain from H&E and IHC images using ImageJ, and applying data augmentation techniques such as flipping, rotation, and adjusting brightness and contrast to address class imbalance.

• Implement prevalent convolutional neural network (CNN) and Transformer models, as described in Table 4 and Table 5 of Ref. [16], and conduct training and inference of these models using Python, preferably with PyTorch.

• Validate the segmentation results predicted by these models across various breast cancer datasets, including H&E and IHC images, by comparing them to the ground truth segmentation mask (e.g., on the MHCI and BCSS datasets) or the ground truth cellularity (e.g., on the BreastPathQ and Post-NAT-BRCA datasets).

• [Optional] Develop multiple instance learning (MIL) techniques to improve segmentation performance across diverse breast cancer datasets, aiming to achieve accuracy comparable to that of supervised semantic segmentation methods.

Resources

• BreastPathQ dataset: a public dataset consisting of 69 H&E stained WSI collected from the resection specimens of 37 post-neoadjuvant therapy patients with invasive residual breast cancer. 2579 image patches with ROI of 512 × 512 pixels are manually annotated with estimated cellularity ranging between [0, 1].

• Other public datasets: https://github.com/maduc7/Histopathology-Datasets

• IHC datasets in NEOCHECKRAY. There are 109 IHC patches stained with an MHC-I antibody with pixel-level manual annotations.

Prerequisite

• Python

Contact persons

Dr. Ir. Jennifer Dhont (jennifer.dhont@hubruxelles.be), Head of Data Science & AI Research Unit at Hopital Universitaire de Bruxelles (Erasme campus)

Pr O. Debeir (olivier.debeir@ulb.be)

references

[1] D. Yan, X. Ju, et al., “Tumour stroma ratio is a potential predictor for 5-year disease-free survival in breast cancer,” BMC Cancer, vol. 22, no. 1, p. 1082, Oct. 2022.

[2] L. Priya C V, B. V G, V. B R, and S. Ramachandran, “Deep learning approaches for breast cancer detection in histopathology images: A review,” Cancer Biomarkers, vol. 40, no. 1, pp. 1–25, May 2024.

[3] M. Peikari, S. Salama, et al., “Automatic Cellularity Assessment from Post-Treated Breast Surgical Specimens,” Cytometry A, vol. 91, no. 11, pp. 1078–1087, Nov. 2017.

[4] S. Akbar, M. Peikari, et al., “Automated and Manual Quantification of Tumour Cellularity in Digital Slides for Tumour Burden Assessment,” Sci Rep, vol. 9, no. 1, p. 14099, Oct. 2019.

[5] X. Catteau, E. Zindy, et al., “Comparison Between Manual and Automated Assessment of Ki-67 in Breast Carcinoma: Test of a Simple Method in Daily Practice,” Technol Cancer Res Treat, vol. 22, p. 15330338231169603, Jan. 2023.

[6] E. H. Allott, S. M. Cohen, et al., “Performance of Three-Biomarker Immunohistochemistry for Intrinsic Breast Cancer Subtyping in the AMBER Consortium,” Cancer Epidemiology, Biomarkers & Prevention, vol. 25, no. 3, pp. 470–478, Mar. 2016.

[7] T. Vougiouklakis, B. J. Belovarac, et al., “The diagnostic utility of EZH2 H-score and Ki-67 index in non-invasive breast apocrine lesions,” Pathology - Research and Practice, vol. 216, no. 9, p. 153041, Sep. 2020.

attached pdf document

Promotor, co-promotor, advisor : olivier.debeir@ulb.be, Julien.Cabay@ulb.be, Thomas.Vandamme@ulb.be

Research Unit : LISA-IMAGE

Description

RAG (Retrieval-Augmented Generation) for Patents

This project consists in the design, development, and testing of a RAG system (an AI chatbot with a specific knowledge base) for a dataset of patents.

Context

Patents are an invaluable economic asset, enabling inventors to protect their inventions for a set duration of time. Those invaluable assets, in the form of patent documents, represent an enormous challenge for the administrations responsible with the protection processes (i.e. Intellectual Property Offices). Those documents are highly technical, composed of different modalities (text and schematics), and are particularly numerous (there were more than 35 million patents in force worldwide as of 2023, source WIPO statistics database).

Recent technological advancements in the field of Artificial Intelligence (AI), namely Large Language Models (LLMs) and the chatbots that these power, carry enormous promises of automation for these complex tasks. One of those, Retrieval-Augmented Generation, is frequently branded as a solution to hallucination in LLMs, as well as enabling a relatively easy specialization of the model using a knowledge library.

Objective

In this project, you will design, develop and test such a solution on a large corpus of patents.

Methods

Different open-source LLMs can be used and benchmarked, as well as the different RAG techniques. The dataset can be sourced from Google Patents.

Prerequisite

• Python
• Machine Learning / Deep Learning

Contact person

For more information please contact : Thomas.Vandamme@ulb.be

Promotor, co-promotor, advisor : olivier.debeir@ulb.be, Julien.Cabay@ulb.be, Thomas.Vandamme@ulb.be

Research Unit : LISA-IMAGE

Description

Design and Implementation of a viewer for IP (Intellectual Property) datasets

This project consists in the design, development, and implementation of a viewer website/software for IP datasets (Trade Marks, Patents, ...). The viewer will enable users, developers and researchers to search, label and extract different relevant aspects of the datasets.

Context

Current dataset viewer tools, such as label studio (https://labelstud.io/), have demonstrated their relevance in research and development ecosystems, especially those related to data and deep learning. However, those tools are not perfect, and several panes of datasets, such as those related to the legal field (esp. text documents) are left out of such solutions.

Objective

In this project, you will develop a complete tool (ideally web-based), or an open-source plug-in for another viewer/labelizer (such as label studio, for example), capable of handling multimodal informations, such as those relative to IP (e.g. images, 3D volumes, schematics, text, sound, ...).

Prerequisite

• Web Technologies
• Python

Contact person

For more information please contact : Thomas.Vandamme@ulb.be

Promotor, co-promotor, advisor : olivier.debeir@ulb.be, Julien.Cabay@ulb.be, Thomas.Vandamme@ulb.be

Research Unit : LISA-IMAGE

Description

Automated web scraping for dataset compilations

This project consists in the design, development, implementation and testing of a series of automated web scrapers. The ultimate goal is to develop a series of tools to enable the acquisition and synchronisation of different datasets.

Context

Deep Learning relies on voluminous and (ideally) good-quality datasets. Those are, unfortunately, hard to gather and label.

In the field of Intellectual Property (IP, including, e.g. Patents, Trade Marks, Designs), some relevant informations are curated by public IP offices (tasked with the administration of the different associated rights; registration, protection, ...). Those public bodies make publicly available a series of information through various search engines (for example, see https://www.euipo.europa.eu/en/search and https://ipportal.wipo.int/home). The offices do not allow for bulk download, and curating by hand these datasets is a particularly tedious task (there are millions of registered rights, for example).

Objective

In this project, you will develop tools to enable the fast development of web scrapers, implementing various measures to disable anti-scraping protections on the websites. A new, untested use case will be chosen to illustrate the tools capabilities.

Methods

You will create web-based applications, interface with (simple) databases and external providers if needed. Your end-product will enable non-developers to choose the elements they want to retrieve automatically in a webpage, and various other settings.

Prerequisite

• Web Technologies
• Python
• (optional) Selenium or other automation software

Contact person

For more information please contact : Thomas.Vandamme@ulb.be

Promotor, co-promotor, advisor : francois.horlin@ulb.be, Jean-Francois Determe,

Research Unit : OPERA - WIRELESS

Description

Context:

Wi-Fi modems are continuously evolving to meet the ever-increasing expectations of the users in terms of communications rates. The last amendment of the 802.11 standard, the 11be amendment referred to as extremely high throughput (EHT), specifies the Wi-Fi communications at frequencies below 7 GHz on a 320 MHz-wide bandwidth and by using up to 16 spatial streams created by arrays of antennas. Communication rates up to 50 Gbps will be supported by your modem!

At the same time, the new amendment 11bf is also developed to support Wi-Fi sensing besides communications. The principle is to leverage the channel measurements already useful for the communications and characterizing the environment to offer new breakthrough applications aiming at improving the autonomy and the security at home. Wi-Fi sensing can be seen as a follow-up of the effort on the design of Wi-Fi based passive radars, i.e. radars working by observing the Wi-Fi communications signals of opportunity to sense the environment.

Objective:

Among the envisioned Wi-Fi sensing use cases, the monitoring of activities to assess the health state or the fast detection of specific events like a fall to quickly call emergency services may help developing the autonomous living of the elderly at home. This calls for the estimation of parameters like the walking speed and the deployment of classification algorithms to detect specific events. The goal of this master thesis is to design a Wi-Fi based passive radar to track people indoors and extract micro-Doppler signatures useful to detect events like a fall or other activities when inputted to a classification algorithm.

Methodology and tools: • Design of Wi-Fi based radar system People tracking indoors based on range/Doppler/angle estimations Micro-Doppler signature extraction based on time-frequency analysis • Settle hardware setup and acquire experimental data (USRP X310/X410) • Conceive a Wi-Fi based health monitoring system Physical parameter estimation to quantify level of activity (actimetry) Classification algorithms for fall detection incl. feature extraction

Contact: francois.horlin@ulb.be

attached pdf document

Promotor, co-promotor, advisor : bram.vanderborght@vub.be, - , Pasquale Ferrentino

Research Unit : R&MM

Description

Project title

The project aims to solve an open issue in a certain domain of application.

Context

The project is done in collaboration with This Company.

Many data are already available

Objective

Here the goal

Methods

Different methods are to be tested...[Ronneberger2015]

Prerequisite

• C,
• C++,
• Python

Contact person

For more information please contact : contact@ulb.be

references

Ronneberger2015

attached pdf document

Promotor, co-promotor, advisor : ilias.el.makrini@vub.be, - , Tom Turcksin

Research Unit : R&MM

Description

Advanced Calibration Methods for EMG: Developing a Customizable Calibration Framework

The project aims to improve the accuracy and relevance of EMG-based assessments by developing a flexible, task-specific calibration framework tailored to human movement analysis.

Context

The project is done in collaboration with AugmentX (VUB). It leverages Cometa and TMSI EMG sensors available at the AugmentX infrastructure (www.augmentx.be). Many data are already available to support development and validation.

Objective

To develop a customizable EMG calibration and normalization framework that offers experimenters multiple task-relevant calibration methods. The focus will be on the elbow joint, incorporating muscle fatigue and joint angle considerations, with the ultimate goal of estimating joint torque from EMG data.

Methods

Different methods are to be tested and developed, namely:

• Literature review of existing EMG calibration techniques;

• Selection and implementation of task-relevant methods;

• Integration of muscle fatigue and joint kinematics into calibration;

• Development of a GUI or script to guide users through calibration steps;

• Testing and validation using Cometa and TMSI sensors.

Prerequisite

• Matlab,
• Python or C++

Contact person

For more information please contact : tom.turcksin@vub.be

Promotor, co-promotor, advisor : tom.verstraten@vub.be, - , M Wu

Research Unit : ROBOTICS AND MULTIBODY MECHANICS RESEARCH GROUP

Description

Context

One of the most critical challenges for development of exoskeletons and exosuits is the design of physical attachments – mechanical straps, cuffs, or sleeves – that connect the robot to the human user. In addition to creating a mechanical connection between the exoskeleton/exosuit to the human body, physical attachments also provide haptic – kinesthetic and proprioceptive – information to human skin, soft tissues, and joints near the attachment site. This haptic (touch-related) information is uncontrolled by the exoskeleton/exosuit and may have unintentional effects on human control of their own movement and muscle activation. The sensorimotor effects of physical attachments have not been little studied, and greater scientific understanding is important for informing future designs of physical attachments for safe and effective exoskeletons/exosuits.

Objective

The student will characterize effects of physical attachments on human movement and muscle activation. This characterization will demonstrate how existing exoskeletons/exosuits may unintentionally influence human behavior through haptic feedback.

Methods

The student will conduct human-subject experiments on unimpaired young adults to measure how different physical attachment types and donning procedures affect human movement and muscle activation patterns. The student will operate motion capture camera and EMG sensor systems to collect data on movement and muscle activity, respectively. Finally, the student will analyze this data in Matlab to extract biomechanics metrics such as joint range of movement and muscle activation level.

Prerequisites

Experience in or willingness to learn human-subject testing Matlab programming for data analysis Experience or interest in studying human biomechanics and sensorimotor control

Contact person

Mengnan.wu@vub.be

Promotor, co-promotor, advisor : michel.kinnaert@ulb.be, - , Louise Massager

Research Unit : SAAS

Description

Context

Europe is developing reusable launchers in the framework of different projects. To ensure launcher reliability, it is necessary to evaluate the state of health of the different parts of the launcher once it comes back to earth without dismantling it. The research project in which this master thesis takes place addresses the development of a systematic methodology to achieve this goal for the electromechanical actuators (EMAs) used to orientate the nozzle and the fins notably.

A model-based simulator of an electromechanical actuator (EMA) is currently used to complement the available experimental dataset (mostly in healthy operation) with synthetic data (of both healthy and faulty operating modes). There is however a mismatch between experimental data and synthetic data (i.e. generated by the simulator). Indeed, the simulator is based on simplified models of healthy operation and faults.

Objective

The goal is to improve the current model-based simulator to generate more realistic data (ultimately to better train the health monitoring system).

Methods

The main idea of the project is to achieve this goal by integrating a data-driven approach to the current simulator[1] in order to generate more realistic synthetic data (i.e. more similar to the limited experimental dataset we currently have and ideally to future experimental data). This way, future experimental data on natural degradation, defects and parameter variations (due to production variability, temperature change, etc.) could be exploited. The data-driven approach could for example be based on an adversarial machine learning approach [2], and on neural networks while enforcing adequate closed-loop performance [3].

The core challenges of the project lie in the limited available experimental dataset and in mitigating the impact of the data-driven approach. The latter point is essential for aerospace applications. Indeed the data-driven approach must not completely alter the physics-based models as the simulator must still be based on physics for ensuring explainability. This stems from the need for certification of the approach, which means that interpretability of how the method operates is required.

The work includes the following steps: 1. perform a bibliographic search on data-driven techniques for synthetic data generation, ideally for hybrid digital twins (i.e. digital twins based on both physics and experimental data). 2. study the EMA models in healthy and faulty states and get acquainted with the current simulator in MATLAB/Simulink 3. process the experimental measurements that will be provided to her/him in order to determine the most relevant features and/or identify parameters to update and relevant constraints 4. compare the most promising data-driven approach for this application in terms of performance and ease of updating once new data are available.

Prerequisite

*Good acquaintance with MATLAB/SIMULINK *Mastery of the bases of electrical machines and control theory

Contact person

For more information please contact : louise.massager@ulb.be and michel.kinnaert@ulb.be

references

[1] Massager, Louise, and Michel Kinnaert. "Modelling of electromechanical actuators in reusable launchers for health monitoring purposes." Internal report, 2024. [2] Goodfellow, Ian J., et al. "Generative adversarial nets." Advances in neural information processing systems 27 (2014). [3] Banderchuk, Ana, Daniel Coutinho, and Eduardo Camponogara. "Combining robust control and machine learning for uncertain nonlinear systems subject to persistent disturbances." 2023 62nd IEEE Conference on Decision and Control (CDC). IEEE, 2023.

attached pdf document

Promotor, co-promotor, advisor : michel.kinnaert@ulb.be, Laurent Catoire,

Research Unit : SAAS

Description

Context

Many of the pilot processes used in the SAAS department to teach control theory were developed through master's theses. This is the case for the rotary inverted pendulum, the ring positioner, the ball in the tube process, …

Objective

The aim of this master thesis is to develop new pilot processes that are modular, evolving, and open-source to provide a better and more practical learning experience to the students. Here are a few examples of processes that SAAS would like to develop (non-exhaustive list): - Ball in hoop or Flying ball in hoop o https://www.youtube.com/watch?v=8FaNk6C2ckM o https://www.youtube.com/watch?v=484GN4KBQnc o https://github.com/aa4cc/flying-ball-in-hoop o https://aa4cc.github.io/flying-ball-in-hoop/ - Cubli - robot that can jump up and balance on its corner o https://www.wevolver.com/specs/cubli - …

Methods

The main steps of the work are:

o selection of the sensors/actuators o design of the signal conditioning / acquisition stages o design of the experimental setup (SolidWorks, 3D printer …) o design of the power supply & cable management o modeling of the process o implementation of a control strategy (Arduino/C programming or Matlab/data-acquisition board or Raspberry PI) o setup of some didactic experiments & their related teaching materials

Prerequisite

• quick & autonomous learner in a dynamic environment
• team player, creativity
• basic knowledge in control theory, digital signal processing, electronics

Contact person

For more information please contact : laurent.catoire@ulb.be and michel.kinnaert@ulb.be

Reference

Xavier Jordens, Robin Wilmart, Emanuele Garone, Michel Kinnaert, Laurent Catoire. A Project-Based Learning Approach for Building an Affordable Control Teaching Lab: The Centrifugal Ring Positionner, IEEE Access, vol 10, pp 4907 – 4918, 2022

attached pdf document

Promotor, co-promotor, advisor : michel.kinnaert@ulb.be, - , Maxime Bussios

Research Unit : SAAS

Description

Context

Renewable energies and electric transportation are the cornerstones for developing a sustainable future society. Energy storage is fundamental in this context, in order to store surplus of energy and use it when the wind does not blow or the sun does not shine, or to produce vehicles that do not pollute the environment when they are on the roads. Among the possibilities, lithium-ion batteries are the technology of choice given their high energy capacity and efficiency. However in contrast with other battery technologies, the benefits of lithium-ion batteries come at the price of careful monitoring requirements. Indeed, faulty cells in a battery pack can have catastrophic consequences including fire.

Objective

The objective of this thesis is to develop a monitoring system that is able to detect and localize the degraded or weak cells within a pack on the basis of available voltage, current and temperature measurements. Both synthetic data obtained from a realistic battery pack simulator, and real data recorded on a 4-cell battery pack will be exploited to determine features that can be extracted from the measurements, or from combinations of measurements, and that exhibit pack malfunction. Next, appropriate classification tools will be investigated in order to decide on the healthy or degraded state of the pack and to localize the degraded cell/cells by processing the features extracted from the measurements. Various degradation levels and types will be considered in order to characterize the sensitivity to each fault.

Methods

The work can be separated in the following tasks: 1. perform a bibliographic search on fault/degradation diagnosis for battery packs, 2. generate synthetic data for heathy pack operation and for various degradation types and levels, 3. Use measurements and/or appropriate functions of the measurements to generate features that exhibit faulty/degraded behaviour, 4. Develop a classification method that decides on the pack state by processing the features extracted from regular measurements.

Requested skills

• quick & autonomous learner in a dynamic environment,
• team player, creativity,
• good knowledge of system and control theory,
• good acquaintance with MATLAB/SIMULINK

Contact person

For more information please contact : maxime.bussios@ulb.be and michel.kinnaert@ulb.be

attached pdf document

Promotor, co-promotor, advisor : michel.kinnaert@ulb.be, - , Louise Massager

Research Unit : SAAS

Description

Context

Europe is developing reusable launchers in the framework of different projects. To ensure launcher reliability, it is necessary to evaluate the state of health of the different parts of the launcher once it comes back to earth without dismantling it. This master thesis should contribute to a research project that addresses the development of a systematic methodology for the health monitoring of the electromechanical actuators (EMAs) used to orientate the nozzle and the fins notably. A classifier for detecting and identifying defects in an electromechanical actuator (EMA) is currently used based mostly on physical models of EMAs and common faults. The current classifier has several limitations which the present master thesis is aimed at overcoming.

Objective

The goal is to improve the current model-based fault detection and isolation (FDI) system. While additional ideas are welcome, different sub-goals have already been identified: • Automated adaptation to new data by incremental learning (no need for re-training from the complete database) • Automated definition of a new type of fault (i.e. generation of a new class) • Provide explanations on the decision taken (i.e. choice of classification; likelihood or belief associated to each class) • Account for the unbalanced database (many more data in healthy mode than in the different faulty modes) An existing classifier based on support vector machines is available[1] and the possibility to equip this type of classifier with the above properties should be evaluated. Possibly other types of classifiers could also be investigated if they offer more flexibility for achieving the indicated goals. Work on this topic has been ongoing for quite some time, notably for image processing[3] [4] [5]. In the present case, it is also important to come up with a solution with low computing cost and to provide an explanation for the decision taken[2]. Those constraints are essential as the approach is aimed for aerospace applications. It must therefore guarantee the performance of the classifier even after the inclusion of new data. The need for certification of the approach should be kept in mind, which means that interpretability of how the method operates is required.

Methods

The project is made of the following steps: 1. perform a bibliographic search on classifiers with adaptability to new data (incremental learning ability) and/or explainability features. 2. study the EMA models in healthy and faulty states and get acquainted with the current simulator in MATLAB/Simulink and FDI system [6]. 3. Get acquainted with the methodology to generate fault indicators and the associated features 4. Compare the most promising data-driven approach for this application in terms of performance and learning ability

Requested skills

• Quick & autonomous learner in a dynamic environment,
• Team player, creativity
• Good acquaintance with MATLAB/SIMULINK
• Good mastery of the bases in in control theory and digital signal processing

Contact person

For more information please contact : louise.massager@ulb.be and michel.kinnaert@ulb.be

references

[1] Wauthion, Benjamin, et al. "Monitoring based on analytical redundancy and classification for a primary flight surface electromechanical actuator." IFAC-PapersOnLine 55.6 (2022): 790-796 [2] Carlevaro, Alberto, et al. "Probabilistic safety regions via finite families of scalable classifiers." arXiv preprint arXiv:2309.04627 (2023). [3] Michael D. Muhlbaier, Apostolos Topalis, and Robi Polikar, Learn++.NC: Combining Ensemble of Classifiers With Dynamically Weighted Consult-and-Vote for Efficient Incremental Learning of New Classes, IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 20, NO. 1 (2009). [4] Liu Yu, Sarah Parisot, Gregory Slabaugh, Xu Jia and Ales Leonardis and Tinne Tuytelaars. More Classifiers, Less Forgetting: A Generic Multi-classifier Paradigm for Incremental Learning, European Conference on Computer Vision (ECCV), 2020 [5] Da-Wei Zhou , Yang Yang and De-Chuan Zhan, Learning to Classify With Incremental New Class, IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, VOL. 33, NO. 6 (2022). [6] Massager, Louise, Geoffrey Postal, and Michel Kinnaert. "Three-phase motor drive fault detection and isolation based on multi-model dual extended Kalman filtering." Benelux Meeting on Systems and Control, 2025.

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, - , Adam CHAFAI adam.chafai@ulb.be

Research Unit : TIPS

Description

Context: Percipio Robotics is a spin-off from the FEMTO-ST research institute, which has designed the Tulip gripper [1]. This compact, lightweight gripper, weighing less than 30g, is designed for micromanipulation and can grip objects from 50µm to 10mm. It solves the problems of large grippers and fragility frequently encountered in micro-robotics. Parallely, the TIPs department designs and manufactures compliant mechanisms in glass (FemtoPRINT technique), whose deformation is measured with optical/photonics techniques.

Objectives: This thesis aims to design and develop an instrumented version of the Percipio Robotics’ Tulip gripper. The master thesis can be preceded by a 3 months internship in the company (Besançon, France).

Methods: Literature review. Functional analysis and requirements. Design. Fabrication and characterization of the flexure mechanism.

Prerequisites: mechanical design, good command of French

Contact: adam.chafai@ulb.be

References: [1] https://ephj.ch/en/percipio-robotics-tulip-gripper-takes-micro-manipulation-to-the-next-level/ [2] L. Amez-Droz et Al. Instrumented Flexible Glass Structure: A Bragg Grating Inscribed with Femtosecond Laser Used as a Bending Sensor, MDPI Sensors, 23, 8018 (2023) [3] M. Tunon de Lara, Femtosecond pulse laser-engineered glass flexible structures instru-mented with an in-built Bragg grating sensor, Optics Express, https://opg.optica.org/oe/fulltext.cfm?uri=oe-31-18-29730&id=536683 (2023)

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, - , Manon CASSIGNOL (manon.nicole.francoise.cassignol@ulb.be)

Research Unit : TIPS

Description

Context: Soft matter can serve as an actuator in microrobotics by deforming under external stimuli such as light, heat, or pH, producing mechanical outputs like force or displacement. At the microscale, these smart materials can be 3D printed without assembly. In our lab, we use two-photon polymerization (2PP) to fabricate soft actuators from a thermo-responsive pol-ymer, poly(N-isopropylacrylamide) (pNIPAM). This material swells below its lower critical solution temperature (LCST) by absorbing water and shrinks above the LCST by expelling it. Recently, we fabricated 50 µm × 50 µm × 50 µm active cubes capable of bending, contract-ing, twisting, or shearing in heated water [1]. To achieve precise, multidirectional motion control, multiple actuators could be combined and selectively triggered by different wave-lengths of light. This is possible by doping them with photothermal nanomaterials that lo-cally convert light into heat [2]. Metallic nanostructures like gold (Au) and silver (Ag) nano-particles or nanorods have been used to actuate PNIPAM-based hydrogels [3]. However, they are usually dispersed uniformly, preventing spatial control. An alternative approach uses a tightly focused femtosecond laser in a PNIPAM hydrogel swollen with silver nitrate, locally forming Ag nanoparticles by multiphoton reduction [4]. Applying this method to our actuators would enable spatially selective nanoparticle patterning, allowing localized, pre-cise activation. Objective: The aim of this thesis is 3D print photosensitive nanoparticles (from a silver nitrate solu-tion) inside PNIPAM hydrogels with the 2PP machine. After printing, light will be used to il-luminate the actuators and will be converted into heat by the nanoparticles. The generated heat will trigger actuator motion by shrinking the hydrogel.

Methods: Literature review. Hydrogel fabrication (with 2PP printing or UV light). 2PP printing of Ag nanoparticles i.e., tune the printing parameters to obtain nanoparticles and optimize the actuation and mechanical properties, print complex deformation structures. Characteriza-tion: absorbance spectra, imaging the nanoparticles, and measuring the light responsive-ness of the structures.

Prerequisites: Materials (to develop the fabrication process and understand the behavior of the hydrogels with and without nanoparticles).

References: [1] Decroly, Gilles, Adam Chafaï, Guillaume de Timary, Gabriele Gandolfo, Alain Delchambre, et Pierre Lambert. 2023. « A Voxel‐Based Approach for the Generation of Advanced Kine-matics at the Microscale ». Advanced Intelligent Systems. [2] Cui, Ximin, Qifeng Ruan, Xiaolu Zhuo, Xinyue Xia, Jingtian Hu, Runfang Fu, Yang Li, Jianfang Wang, et Hongxing Xu. 2023. « Photothermal Nanomaterials: A Powerful Light-to-Heat Con-verter ». Chemical Reviews 123 (11): 6891 6952. [3] Park, Daehwan, Jin Woong Kim, et Chinedum O Osuji. 2024. « Programmable Thermo- and Light-Responsive Hydrogel Actuator Reinforced with Bacterial Cellulose ». [4] Nishiyama, Hiroaki, Shun Odashima, et Suguru Asoh. 2020. « Femtosecond Laser Writing of Plasmonic Nanoparticles inside PNIPAM Microgels for Light-Driven 3D Soft Actuators ». Op-tics Express 28 (18): 26470 80.

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, - , Manon CASSIGNOL (manon.nicole.francoise.cassignol@ulb.be)

Research Unit : TIPS

Description

Context: Soft matter is used as an actuator in microrobotics. It can deform under an external stimulus such as light, heat, or pH to generate a mechanical output (force and displace-ment). At the microscale, these smart materials can be 3D printed without assembly. In the lab, we use the two-photon polymerization method (2PP) to shape 50µm soft actuators out of a thermo-responsive polymer (pNipam = poly(N-isopropylacrylamide)). These active cu-bes demonstrate bending, contraction, twist, or shear deformation in a heated water bath [1]. Their mechanical performances such as Young modulus, force-displacement character-istics, or response time must now be characterized.

Objective: The aim of this thesis is to develop a setup to measure the force-displacement character-istics of such actuators. Inspired by Micro-Electro-Mechanical Systems (MEMS) force sen-sors [2] and/or atomic force microscopy (AFM) [3], this set-up will be fabricated in using glass microstructures (to be produced with the FemtoPrint machine) or with other materials deemed relevant by the candidate.

Methods: Literature review on characterizing the mechanical performance of soft material at mi-croscale. Select the most suitable device. Design the set-up considering the following crite-ria: 1) samples are characterized in water to allow them to swell and shrink, 2) a heating system (conventional or laser) will be used to drive the actuators, and 3) the sensor must be in contact with small samples (50 to 200 µm). Eventually, the results obtained may be supplemented and compared with data obtained with an environmental AFM, at UMons, and/or a nanoindentation system [4], at EMPA (Thun, Switzerland).

Prerequisites: Mechanics (to determine the device shape and develop the different part of the set-up using CAD software), coding (to automatically control the setup), and materials (to un-derstand the material model obtained from experimental measurements).

References: [1] G. Decroly, A. Chafaï, G. de Timary, G. Gandolfo, A. Delchambre, et P. Lambert, « A Voxel‐Based Approach for the Generation of Advanced Kinematics at the Microscale », Ad-vanced Intelligent Systems, 2023, doi: 10.1002/aisy.202200394. [2] M. Lamba, N. Mittal, K. Singh, et H. Chaudhary, « Design analysis of polysilicon piezoresis-tors PDMS (Polydimethylsiloxane) microcantilever based MEMS Force sensor », Int. J. Mod. Phys. B, vol. 34, no 09, p. 2050072, avr. 2020, doi: 10.1142/S0217979220500721. [3] A. Chau, S. Régnier, A. Delchambre, et P. Lambert, « Theoretical and Experimental Study of the Influence of AFM Tip Geometry and Orientation on Capillary Force », Journal of Adhe-sion Science and Technology, vol. 24, no 15 16, p. 2499 2510, janv. 2010, doi: 10.1163/016942410X508307. [4] T. Spratte et al., « Increasing the Efficiency of Thermoresponsive Actuation at the Mi-croscale by Direct Laser Writing of pNIPAM », Advanced Materials Technologies, vol. 8, no 1, p. 2200714, 2023, doi: 10.1002/admt.202200714.

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, Benoit HAUT, Clément RIGAUT

Research Unit : TIPS

Description

Context: The nose is responsible for heating and humidifying the air entering the respiratory tract. While it is only around 10 cm long, it can bring ambient air to a temperature of about 30°C in the pharynx. This function of conditioning the air before reaching the lower respiratory tract is vital to avoid inflammation, asthma and increased risk of infections. Despite the im-portance of this function of the nose, the heating of the air in the nasal cavity remains largely unknown.

Objective: This thesis aims to compute the temperature of the air exiting the nose under various conditions (rest, light effort, moderate effort,...) and ambient temperatures. The condition-ing efficiency of different noses can be compared to deduce the influence of anatomical features on air conditioning.

Methods: First, the simulation models will be created from STL files of nasal cavities. Then simula-tions will be carried out using OpenFOAM software to measure the temperature of the air exiting the nose for various parameters. Finally, the results of the different anatomies will be compared to extract the anatomical characteristics impacting air conditioning.

Prerequisites: • Fluid Dynamics • Thermodynamics

Contact: Clément Rigaut (clement.rigaut@ulb.be)

References: [1] D.-W. Kim, S.-K. Chung, et Y. Na, « Numerical study on the air conditioning characteristics of the human nasal cavity », Computers in Biology and Medicine, vol. 86, p. 18 30, juill. 2017, doi: 10.1016/j.compbiomed.2017.04.018. [2] S. Naftali, M. Rosenfeld, M. Wolf, et D. Elad, « The Air-Conditioning Capacity of the Human Nose », Ann Biomed Eng, vol. 33, nᵒ 4, p. 545 553, avr. 2005, doi: 10.1007/s10439-005-2513-4.  

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, Benoît HAUT, Clément RIGAUT

Research Unit : TIPS

Description

Context: While nasal drugs have been widely used to treat local symptoms of colds or allergies, they have more recently emerged as a potential method for delivering neurological drugs. Indeed, the nose is highly vascularized, which ensures that molecules deposited in the nasal cavity will be readily absorbed into the bloodstream. Moreover, there is growing evidence that drugs can pass directly from the nose to the brain via the olfactory nerve [1]. However, a major drawback of nasal administration is its strong dependence on the individual’s anatomy [2]. Therefore, personalized models are needed to predict the outcomes of nose-to-brain treatments.

Objective: This thesis aims to develop a pharmacokinetic model for the nasal administration of a neurological drug. The model will take into account the location of drug deposition (which varies between individuals), the transfer of the drug to the blood and brain, and its subse-quent elimination. The goal is to determine the optimal treatment plan for each patient (e.g., one large dose or multiple smaller doses, liquid or dry spray, etc.) and predict the in-dividual outcomes of these treatments.

Methods: Based on an extensive literature review, a pharmacokinetic model of the drug will be developed. Existing 3D-printed nasal replicas will be used to assess the distribution of the spray within the nasal cavity. These experimental data will allow for predictions of the treatment outcomes for a given anatomy and help identify the most suitable therapeutic approach for each individual.

Prerequisites: • Knowledge of a programming language

Contact: Clément Rigaut (clement.rigaut@ulb.be)

References: [1] L. Illum, ‘Is nose-to-brain transport of drugs in man a reality?’, Journal of Pharmacy and Pharma-cology, vol. 56, no. 1, pp. 3–17, Jan. 2004, doi: 10.1211/0022357022539. [2] C. Rigaut et al., ‘What Are the Key Anatomical Features for the Success of Nose-to-Brain Delivery? A Study of Powder Deposition in 3D-Printed Nasal Casts’, Pharmaceutics, vol. 15, no. 12, p. 2661, Nov. 2023, doi: 10.3390/pharmaceutics15122661.  

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, - , Margaux MANNNAERTS

Research Unit : TIPS

Description

Context: Lung cancer is the leading cause of cancer death worldwide [1]. As part of the screening process, lung nodules (suspected cancer) are regularly found in peripheral areas that are difficult to access by endoscopy. As most of these nodules are not cancerous, it is essential to be able to take a local biopsy to make a precise diagnosis. However, the lung is like a lab-yrinth, with sections that shrink with each division, and access to a precise peripheral zone is difficult. In addition, the need to use flexible and miniaturized tools implies certain limita-tions. Indeed, the need for flexibility is necessary to avoid damaging the tissue or injuring the patient but means that the tools may deform before the biopsy is taken.

A family of solutions that are being developed uses the concept of controllable/variable stiffness to cope with these issues [2]. These solutions use materials and/or specific geometries that can change rigidity given a certain stimuli (change of temperature, pressure, …).

Objectives: Develop a prototype of a variable stiffness catheter using different equipment present in the lab (molding techniques, 3D printers).

Methods: Literature review. Functional analysis and requirements. Design. Fabrication and evalua-tion of the built prototype.

Prerequisites: • Mechanical design • Interest for mechanical and biomedical engineering

Contact: margaux.mannaerts@ulb.be

References: [1] Global Burden of Disease 2019 Cancer Collaboration et al., « Cancer Incidence, Mortality, Years of Life Lost, Years Lived With Disability, and Disability-Adjusted Life Years for 29 Can-cer Groups From 2010 to 2019: A Systematic Analysis for the Global Burden of Disease Study 2019 », JAMA Oncol., vol. 8, no 3, p. 420, mars 2022, doi: 10.1001/jamaoncol.2021.6987. [2] L. Blanc, A. Delchambre, et P. Lambert, « Flexible Medical Devices: Review of Controllable Stiffness Solutions », Actuators, vol. 6, no 3, p. 23, juill. 2017, doi: 10.3390/act6030023.

attached pdf document

Promotor, co-promotor, advisor : pierre.lambert@ulb.be, - , Margaux MANNNAERTS

Research Unit : TIPS

Description

Context: Lung cancer is the leading cause of cancer death worldwide [1]. As part of the screening process, lung nodules (suspected cancer) are regularly found in peripheral areas that are difficult to access by endoscopy. As most of these nodules are not cancerous, it is essential to be able to take a local biopsy to make a precise diagnosis. However, the lung is like a lab-yrinth, with sections that shrink with each division, and access to a precise peripheral zone is difficult. In addition, the need to use flexible and miniaturised tools implies certain limita-tions. Indeed, the need for flexibility is necessary to avoid damaging the tissue or injuring the patient, but means that the tools may deform before the biopsy is taken. One way to ensure that the biopsy is taken at the right location is to have knowledge on the position and deformation of the catheter tip. Despite the exploration of various technologies such as electromagnetic sensors (EM), optical fibers, X-rays, etc [2], [3] , biopsy outcomes remain highly variable and dependent on a variety of factors including the type and number of used equipment, experience of the practician, location of the nodule in the lung. [4]

Objectives: This master thesis aims to design and develop a system enabling the practicians to know how the tip of the catheter is deformed in the lungs, due to their mechanical contact with the bronchii and the internal efforts developed in the catheter. Given the very small size of the peripheral bronchi (<1 mm), the system can be initially developed at a larger scale. Some inspiration can be taken from textile-based sensors, or other resistive strain gauges [5].

Methods: Literature review. Functional analysis and requirements. Design. Fabrication and charac-terization of a shape sensing catheter tip.

Prerequisites: • Mechanical design, electronics • Interest for mechanical and biomedical engineering

Contact: margaux.mannaerts@ulb.be

References: [1] J. M. Kocarnik et al., “Cancer Incidence, Mortality, Years of Life Lost, Years Lived With Disabil-ity, and Disability-Adjusted Life Years for 29 Cancer Groups From 2010 to 2019 A Systematic Analysis for the Global Burden of Disease Study 2019,” JAMA Oncol, vol. 8, no. 3, pp. 420–444, 2022, doi: 10.1001/jamaoncol.2021.6987. [2] C. Shi et al., “Shape sensing techniques for continuum robots in minimally invasive surgery: A survey,” IEEE Trans Biomed Eng, vol. 64, no. 8, pp. 1665–1678, Aug. 2017, doi: 10.1109/TBME.2016.2622361. [3] R. Brekken et al., “Accuracy of instrument tip position using fiber optic shape sensing for navigated bronchoscopy,” Med Eng Phys, vol. 125, Mar. 2024, doi: 10.1016/j.medengphy.2024.104116 [4] J. Thiboutot et al., “Accuracy of Pulmonary Nodule Sampling Using Robotic Assisted Bron-choscopy with Shape Sensing, Fluoroscopy, and Radial Endobronchial Ultrasound (The AC-CURACY Study),” Respiration, vol. 101, no. 5, pp. 485–493, Mar. 2022, doi: 10.1159/000522514. [5] S. Wu, « A n Overview of Hierarchical Design of Textile-Based Sensor in Wearable Electron-ics », Crystals, vol. 12, no 4, p. 555, avr. 2022, doi: 10.3390/cryst12040555.

attached pdf document

Promotor, co-promotor, advisor : benoit.haut@ulb.be, Pierre Lambert, Clément Rigaut

Research Unit : TRANSFERS, INTERFACES AND PROCESSES

Description

Context: The nose is responsible for heating and humidifying the air entering the respiratory tract. While it is only around 10 cm long, it can bring ambient air to a temperature of about 30°C in the pharynx. This function of conditioning the air before reaching the lower respiratory tract is vital to avoid inflammation, asthma and increased risk of infections. Despite the importance of this function of the nose, the heating of the air in the nasal cavity remains largely unknown.

Objective: This thesis aims to compute the temperature of the air exiting the nose under various conditions (rest, light effort, moderate effort,...) and ambient temperatures. The conditioning efficiency of different noses can be compared to deduce the influence of anatomical features on air conditioning. Methods: First, the simulation models will be created from STL files of nasal cavities. Then simulations will be carried out using OpenFOAM software to measure the temperature of the air exiting the nose for various parameters. Finally, the results of the different anatomies will be compared to extract the anatomical characteristics impacting air conditioning.

Prerequisites: • Fluid Dynamics • Thermodynamics

Contact: Clément Rigaut (clement.rigaut@ulb.be)

References: [1] D.-W. Kim, S.-K. Chung, et Y. Na, « Numerical study on the air conditioning characteristics of the human nasal cavity », Computers in Biology and Medicine, vol. 86, p. 18 30, juill. 2017, doi: 10.1016/j.compbiomed.2017.04.018. [2] S. Naftali, M. Rosenfeld, M. Wolf, et D. Elad, « The Air-Conditioning Capacity of the Human Nose », Ann Biomed Eng, vol. 33, nᵒ 4, p. 545 553, avr. 2005, doi: 10.1007/s10439-005-2513-4.

Promotor, co-promotor, advisor : benoit.haut@ulb.be, Pierre Lambert, Clément Rigaut

Research Unit : TRANSFERS, INTERFACES AND PROCESSES

Description

Context: While nasal drugs have been widely used to treat local symptoms of colds or allergies, they have more recently emerged as a potential method for delivering neurological drugs. Indeed, the nose is highly vascularized, which ensures that molecules deposited in the nasal cavity will be readily absorbed into the bloodstream. Moreover, there is growing evidence that drugs can pass directly from the nose to the brain via the olfactory nerve [1]. However, a major drawback of nasal administration is its strong dependence on the individual’s anatomy [2]. Therefore, personalized models are needed to predict the outcomes of nose-to-brain treatments.

Objective: This thesis aims to develop a pharmacokinetic model for the nasal administration of a neurological drug. The model will take into account the location of drug deposition (which varies between individuals), the transfer of the drug to the blood and brain, and its subsequent elimination. The goal is to determine the optimal treatment plan for each patient (e.g., one large dose or multiple smaller doses, liquid or dry spray, etc.) and predict the individual outcomes of these treatments.

Methods: Based on an extensive literature review, a pharmacokinetic model of the drug will be developed. Existing 3D-printed nasal replicas will be used to assess the distribution of the spray within the nasal cavity. These experimental data will allow for predictions of the treatment outcomes for a given anatomy and help identify the most suitable therapeutic approach for each individual.

Prerequisites: • Knowledge of a programming language

Contact: Clément Rigaut (clement.rigaut@ulb.be)

References: [1] L. Illum, ‘Is nose-to-brain transport of drugs in man a reality?’, Journal of Pharmacy and Pharmacology, vol. 56, no. 1, pp. 3–17, Jan. 2004, doi: 10.1211/0022357022539. [2] C. Rigaut et al., ‘What Are the Key Anatomical Features for the Success of Nose-to-Brain Delivery? A Study of Powder Deposition in 3D-Printed Nasal Casts’, Pharmaceutics, vol. 15, no. 12, p. 2661, Nov. 2023, doi: 10.3390/pharmaceutics15122661.

Promotor, co-promotor, advisor : benoit.haut@ulb.be, Jérémy Rabineau, Clément Rigaut

Research Unit : TRANSFERS, INTERFACES AND PROCESSES

Description

L'objectif de ce mémoire est de développer des outils et des procédures pour quantifier certains paramètres du flux de sang dans le ventricule gauche et l'aorte, en se basant sur des images d'IRM 4D flow.

En se basant sur des outils open-source existants, il s'agira de commencer par calculer quelques métriques basiques : volume d'éjection systolique, débit sanguin maximal, etc. Ensuite, des paramètres plus avancés, liés à la complexité du flux et à l'efficacité de l'activité cardiaque, seront intégrés : vorticité, hélicité, énergie cinétique turbulente, etc.

Ces outils pourront être testés sur des bases de données disponibles : comparaison d'astronautes avant et après vol spatial, comparaison de volontaires sains avant, pendant et après deux mois d'alitement "tête en bas" (simulation de micropesanteur), ou encore comparaison de patients et de volontaires sains, avec classification en fonction de certaines pathologies.

Contact : Benoit Haut

Promotor, co-promotor, advisor : john.lataire@vub.be, ICU director Prof. Joop Jonckheer, Andy Keymolen

Research Unit : VUB-ELEC

Description

Respiratory Oscillometry (RO) - Advancing Respiratory Monitoring: Refining Data-Driven Modelling for Longitudinal Assessment in Clinical Practice.

In this master's thesis project, you will delve into the realm of respiratory monitoring to unearth valuable insights that could improve patient care. The objective of this thesis is to discern statistically significant parameter variations across multiple measurements obtained from individual patients. The proposed approach is as follows: Getting up to topic: • Exploration of Respiratory Oscillometry (RO): Unravel the potential of RO in assessing lung impedance. Through a comprehensive survey of literature and state-of-the-art techniques, you will gain a deep understanding of how RO measurements can illuminate the health status of patients. • Hands-on Simulation and Mc Invent Trial Setup: Gain practical experience by simulating RO techniques in various realistic scenarios. Familiarize yourself with the intricacies of the clinical trial setup and the experiments conducted that gathered the data. • Data Processing Mastery: Learn the art of data cleaning and preprocessing to ensure the accuracy and reliability of measurements. Through meticulous removal of transients and breathing artefacts, you will prepare the data for in-depth analysis. Design and development of novel techniques. • Innovative Parameter Estimation Techniques: Elevate your skills by improving estimation techniques to enable longitudinal assessment of model parameters. Begin with a first-order model and progress towards more sophisticated models tailored to capture diagnostic-rich low-frequency regions [1]. • Efficient Automation: Streamline the analysis process by automating data processing, allowing for efficient examination of multiple patient datasets. • Development of User-Friendly Application: Translate your findings into actionable insights with the development of a user-friendly application. This application will provide clinicians with longitudinal assessments of identified parameters, from the MC Invent patients as an example but applicable in future trials, facilitating informed decision-making in patient care. • Collaboration with Physicians: Engage in discussions with physicians to validate the clinical relevance of identified parameters and ensure the ergonomic design of the application. Your collaboration will bridge the gap between technological innovation and clinical practice. • Predictive Modelling for Liberation from Mechanical Ventilation: Utilize the longitudinal assessments to develop a predictive model for determining a patient's readiness for weaning from mechanical ventilation. This predictive tool will empower clinicians to make informed decisions, ultimately leading to improved patient outcomes. Work approach: We encourage you to take ownership of the project, unleash your creativity, and push the boundaries of oscillometry. With access to dedicated workspace and state-of-the-art equipment, you will have the support and resources needed to excel in your research endeavors.

For more information please contact : John.Lataire@vub.be

References

[1] Bates, Jason HT. Lung mechanics: an inverse modelling approach. Cambridge University Press, 2009

attached pdf document

Promotor, co-promotor, advisor : john.lataire@vub.be, ICU director Prof. Joop Jonckheer, Andy Keymolen

Research Unit : VUB-ELEC

Description

Respiratory Oscillometry (RO) - Advancing Respiratory Mechanics Analysis: A time-varying modelling approach.

In the intricate dance of breathing, the inspiratory phase sees respiratory muscles contract, while during exhalation, they relax. This physiological ballet holds critical clues for clinicians, as contracted muscles typically exhibit increased stiffness. Leveraging this insight, we anticipate periodic fluctuations in the elastance estimate of a patient's respiratory system throughout each breath. For ICU physicians, this dynamic provides invaluable insights, with heightened inspiratory stiffness signalling strengthening respiratory muscles and potentially heralding readiness for liberation from mechanical ventilation. Yet, elastance is just the tip of the iceberg. As we delve deeper into within-breath analysis, we uncover a realm ripe for exploration within respiratory oscillometry. While non-parametric estimations have proven highly beneficial [1], parametric modelling has thus far shown promise primarily in simulation [2]. With this thesis, we endeavour to push the boundaries of innovation by crafting a robust parametric identification strategy for within-breath analysis in mechanically ventilated patients. Our overarching goal is clear: to discern statistically significant within-breath parameter changes across multiple measurements from individual patients. To achieve this, we propose a multifaceted approach:

Design and development of novel techniques. • Implement estimation techniques to detect a time-varying behaviour of the model parameters. Starting from [2] and improving towards more complex models • Design an excitation signal that allows an improved identification of the time-varying parameters • Design an experiment to prove that the excitation signal improves upon the state-of-the-art • Develop a longitudinal assessment strategy for the time-varying parametersWork approach: We encourage you to take ownership of the project, unleash your creativity, and push the boundaries of oscillometry. With access to a dedicated workspace and state-of-the-art equipment, you will have the support and resources needed to excel in your research endeavours.

For more information please contact : John.Lataire@vub.be

References

[1] Veneroni, Chiara, et al. "Oscillatory respiratory mechanics on the first day of life improves prediction of respiratory outcomes in extremely preterm newborns." Pediatric Research 85.3 (2019): 312-317. [2] Alamdari, Hamed Hanafi, Kamal El-Sankary, and Geoffrey N. Maksym. "Time-varying respiratory mechanics as a novel mechanism behind frequency dependence of impedance: a modeling approach." IEEE transactions on biomedical engineering 66.9 (2018): 2433-2446.

attached pdf document

Promotor, co-promotor, advisor : john.lataire@vub.be, ICU director Prof. Joop Jonckheer, Andy Keymolen

Research Unit : VUB-ELEC

Description

Respiratory Oscillometry (RO) - Investigating the Impact of Oscillometry on Ventilation Asynchrony: A Novel Approach

Ventilation asynchrony, a critical concern in patient care, occurs when ventilators fail to detect a patient's inspiratory effort in a timely manner. Traditionally, this effort is identified through peak flow signals generated during inspiration. However, the introduction of oscillometry into ventilation monitoring has introduced new challenges, as it overlays excitation signals onto the ventilation waveform, leading to subtle flow fluctuations during expiration. This master's thesis project aims to explore whether oscillometry exacerbates ventilation asynchrony. Leveraging patient data from the MC Invent trial, in this thesis you will analyze both oscillometry and ventilation measurements to discern any differences in ventilation asynchrony prevalence. Utilizing state-of-the-art identification techniques [1], the project will adapt methodologies to suit the specific context, enabling precise analysis. In scenarios where ventilation asynchrony is induced, the project will propose and validate an excitation signal design strategy aimed at mitigating its prevalence. This strategy will be tested on simulators to ensure efficacy and patient safety. Furthermore, in cases where ventilation asynchrony is not observed, the project will enhance the design of excitation signals. Current approaches often utilize exponential amplitude decay with a fixed peak-to-peak limit, which may not account for individual patient characteristics. By employing a model-based design strategy, the project seeks to optimize signal amplitudes and minimize flow perturbations that disturb the patient, thus improving patient comfort and outcomes.

Design and development of novel techniques. • Implement a ventilation asynchrony detection mechanism starting from [1] • Comparative analysis of ventilation asynchrony prevalence between ventilation and oscillometry measurements. • Development of an innovative excitation signal design strategy tailored to address ventilation asynchrony concerns or improve the amplitude design strategy. • Experimental validation of the proposed excitation signal's efficacy, surpassing existing benchmarks.

Work approach: To facilitate efficient collaboration and feedback, the project offers a dedicated workspace within the department. Access to a professional ventilator with oscillometry functionality enables real-world testing on lung emulators, ensuring the practicality and relevance of developed techniques.

For more information please contact : John.Lataire@vub.be

Reference

[1] L. van de Kamp, J. Reinders, B. Hunnekens et al., Automatic patient-ventilator asynchrony detection framework using objective asynchrony definitions. IFAC Journal of Systems and Control (2024), doi: https://doi.org/10.1016/j.ifacsc.2023.100236

attached pdf document

Promotor, co-promotor, advisor : john.lataire@vub.be, - , Andy Keymolen

Research Unit : VUB-ELEC

Description

Respiratory Oscillometry (RO) - Advancing Respiratory Monitoring: Extending Low-Frequency Oscillometry to Non-Invasive Mask Ventilation

In the Mc-Invent trial, low-frequency RO provided invaluable insights into respiratory mechanics of patients undergoing ventilation via an endotracheal tube (a tube inserted in the trachea). However, as patients transition to mask ventilation post-tube removal, the current monitoring techniques become ineffective, leaving clinicians without vital information crucial for patient care. Your mission: enhance low-frequency RO techniques to enable accurate estimations during mask ventilation. To achieve this, you will delve into the complexities posed by higher flow leakage and dead volume inherent in mask ventilation setups. Your approach will involve: In this thesis, you will improve the low-frequency RO technique such that viable estimations can be obtained during mask ventilation. You will adapt the RO techniques, the excitation signals and modelling processes, such that it can deal with the higher flow leakage (air that leaks via the mask boundaries) and the higher dead volume (air that is in between the sensor and the patient's mouth, encapsulated by the mask). Both elements influence respiratory mechanics estimations negatively. To achieve this, we suggest the following approach:

Design and development of novel techniques. • Understanding the Impact of Dead Volume and Leakage Flow: Through rigorous simulation studies, you will investigate the influence of dead volume and leakage flow on respiratory mechanics estimation. This foundational research will inform subsequent technique development. • Refinement of Modelling Strategies: Armed with insights from simulation studies, you will refine modelling strategies to mitigate the negative impact of dead volume and leakage flow. Incorporating prior knowledge, you will develop robust models capable of accurately capturing respiratory dynamics. • Innovative Excitation Signal Design: Central to your endeavor is the design of an excitation signal strategy that is resilient to the challenges posed by dead volume and leakage flow. Through novel design approaches, you will ensure the reliability and accuracy of respiratory mechanics estimations. • Experimental Validation: Rigorous experimentation is paramount to validate the efficacy of your developed techniques. You will design and conduct experiments to demonstrate the working principle and real-world applicability of your innovations, cementing their status as game-changing advancements in respiratory monitoring.

approach: We encourage you to take ownership of the project, unleash your creativity, and push the boundaries of oscillometry. With access to dedicated workspace and state-of-the-art equipment, you will have the support and resources needed to excel in your research endeavors

For more information please contact : John.Lataire@vub.be

attached pdf document