Below you will find frequently asked questions, divided over four different groups. First, a generic FAQ with information applying to a broad set of master degrees and then more specific FAQs applying to specific programs only.
âSignal Processing in the AI eraâ was the tagline of this yearâs IEEE International Conference on Acoustics, Speech and Signal Processing, taking place in Rhodes, Greece.
In this context, Brent de Weerdt, Xiangyu Yang, Boris Joukovsky, Alex Stergiou and Nikos Deligiannis presented ETRO’s research during poster sessions and oral presentations, with novel ways to process and understand graph, video, and audio data. Nikos Deligiannis chaired a session on Graph Deep Learning, attended the IEEE T-IP Editorial Board Meeting, and had the opportunity to meet with collaborators from the VUB-Duke-Ugent-UCL joint lab.
Featured articles:

Lars Bonnefoy has been awarded an FWO PhD fellowship for his project âFrom Clouds to Kilowatts: Multimodal Deep Learning for Probabilistic Solar Irradiance Nowcasting and Solar Power Forecastingâ. With over 11 GW of solar capacity installed in Belgium, accurate forecasts for the next hours are increasingly important for grid stability. Lars will develop an open-source deep learning system combining satellite imagery, weather model output and ground measurements into probabilistic solar irradiance forecasts. The project is embedded in the Flanders Alliance for Climate Technology (FACT): together with the EVERGi group, the forecasts will be tested at the Smart Village Lab in Research Park Zellik. Lars is supervised by Prof. Lesley De Cruz (VUB-ETRO/RMI), with Prof. Thierry Coosemans (VUB-EVERGi) and Dr. Pierre de Buyl (RMI) as co-supervisors.
Congratulations, Lars!
A free trade agreement (FTA) is a framework. It is essential, but nobody lives in a structure. It still needs walls: institutions and rules companies can rely on. Windows: real market access, in both directions. Light: predictable, transparent conditions. Ventilation: talent and ideas moving freely. And movement: goods, investment and people actually flowing through the building.
That was the thread through the discussion. Beyond tariffs, the conversation kept returning to deeper partnerships in research, technology and innovation, and to the hard part: turning good research into solutions that can be tested, commercialised and scaled to benefit both sides.
The partnership works because each side brings what the other needs. India brings scale, talent and a fast-growing data ecosystem and an ability for rapid deployment. Belgium and the wider European Union bring research excellence and specialised industrial strengths, from life sciences and chemicals to clean technology. Belgium invests 3.4% of its gross domestic product in research and development, among the highest rates in Europe. Through Antwerp-Bruges, one of Europe’s largest ports, it offers a gateway to a single market of 450 million people, and European standards are trusted worldwide. Together, that creates real opportunities for companies on both sides, including small and medium-sized enterprises.
For India, the European Union is already one of its two largest trading partners. For Europe, India is still around 2% of its trade. That gap is the opportunity.
Thank you to Geoffroy Gueyffier , Asim Anwar, Erika Staël von Holstein, Renaat Schrooten , Edith N. NordmannTej Paul Bhatla, Prof. Dr. Johan Stiens ,Hemakiran Gupta , Jan Luykx and Frank Haak for conversations of real substance. And a special thank you to Rakesh Kapila and BICA, the Belgium India Creative Association, for bringing together such a diverse group of voices.
Ambassador Jan Luykx set the bar for success: if, in ten years, India and the European Union are shaping world geopolitics together, we will know the FTA has worked.
Research by Redona Brahimetaj and colleagues at ETRO-imec shows that the tissue surrounding breast microcalcifications can provide more diagnostic information than the calcium deposits themselves, opening new perspectives for breast cancer imaging.
Breast microcalcifications are tiny calcium deposits that can appear in breast tissue and are important indicators in breast cancer screening. Traditionally, imaging techniques have focused on the shape, size and distribution of these deposits to help distinguish benign from malignant lesions.
A new study led by Redona Brahimetaj at the Department of Electronics and Informatics (ETRO-imec) of Vrije Universiteit Brussel challenges this perspective. The research demonstrates that the tissue immediately surrounding individual microcalcifications contains valuable diagnostic information, potentially even more than the calcifications themselves.
Using high-resolution three-dimensional micro-computed tomography (micro-CT), the research team analysed 3,504 individual microcalcifications from 94 breast biopsy specimens, achieving a spatial resolution of approximately 8 micrometres.
By combining advanced medical image analysis, radiomics and machine-learning techniques, the researchers systematically investigated where the most useful diagnostic information originates: within the calcium deposits or in the surrounding breast tissue.
The results were striking. Including the surrounding tissue improved classification performance, with the area under the receiver operating characteristic curve (AUC) increasing from approximately 0.69 to 0.81.
Remarkably, analysing the surrounding tissue alone achieved comparable performance, demonstrating that the microenvironment around the calcifications is a particularly important source of diagnostic information.
This work highlights ETRO’s expertise in medical image processing, quantitative image analysis and artificial intelligence, and illustrates how advanced computational methods can reveal information that conventional imaging analysis may overlook.
By shifting attention from the calcifications themselves to the interface between calcium deposits and surrounding tissue, the research provides a new direction for developing quantitative imaging biomarkers.
Although the study used highly detailed images of biopsy specimens rather than conventional clinical breast images, its findings offer important insights that may guide future breast cancer imaging research. Further validation and investigation of the underlying tissue structures will be necessary before clinical applications can be considered.
The study was conducted through a collaboration between ETRO-imec, VUB’s Laboratory of Experimental Pathology, and the Departments of Pathology and Radiology at UZ Brussel.
The research brings together expertise in artificial intelligence, medical imaging, radiology and pathology, demonstrating the value of interdisciplinary collaboration in addressing challenging medical diagnostic problems.
Micro-CT Radiomics Reveals the Microcalcification-Surrounding Tissue Interface as the Diagnostic Epicenter of Breast Microcalcifications
Authors: Redona Brahimetaj, Jan Cornelis, Guillaume Courtoy, Wim Waelput, Inneke Willekens, Johan De Mey, Ramses Forsyth and Bart Jansen.
https://www.vub.be/en/news/beyond-calcification-vub-imec-researchers-identify
On 30 October 2026 at 16:00, Sebastiaan Priem will defend their PhD entitled âBREATH-BY-BREATH INDIRECT CALORIMETRY BASED ON A DYNAMIC HALDANE TRANSFORMATION: FROM THEORETICAL FRAMEWORK TO CLINICALLY EVALUATED PROTOTYPEâ.
Everybody is invited to attend the presentation in room K.2.56 or online via this link.
Abstract â
Indirect calorimetry is considered the gold standard for measuring resting energy expenditure and respiratory quotient and relies on the measurement of a patient’s rate of oxygen consumption and carbon dioxide production. Despite its established value, its use is constrained by several limitations. First, the use of the Haldane transformation to infer inspired flow, amplifies measurement errors as the inspired oxygen fraction increases, which prevents measurements in patients receiving more than 60-70 % oxygen. Secondly, a single short-term measurement is typically extrapolated to represent the energy expenditure of the patient over a period of 24 hours, which doesn’t capture intra- and inter-day variability. Finally, the cost, form factor, maintenance burden and requirement of trained personnel of available devices confine their use to specialised clinical or research environments.
This dissertation establishes a theoretical framework for breath-by-breath indirect calorimetry that determines the oxygen consumption rate without measuring the expired oxygen fraction and without assuming a value for the respiratory quotient. When both the inspired and expired flow rates are measured, the expired inert gas fraction follows from the mass balance of inert gases, and the expired oxygen fraction from the composition of the expired gas. By applying this method, the inspired oxygen fraction no longer appears in a denominator, which analytically avoids the error amplification at high inspired oxygen fractions. Additionally, a dynamic Haldane transformation is introduced, which distributes the volume correction over the exhalation according to the expired CO2 waveform, and thereby accounts for the dead-space gas at the start of each exhalation.
The requirements for breath-by-breath gas exchange measurements in the clinic and at home are derived from a criterion on the error of REE, and the sensing principles for each quantity are evaluated against them. This analysis shows that flow and CO2 can be measured directly, whereas the expired oxygen fraction cannot, and leads to a compact prototype that measures flow and CO2 at the airway opening.
In a pilot comparison with a clinical reference calorimeter (Q-NRG, COSMED) in 12 healthy volunteers, the agreement for REE and gas exchange rates showed no statistically significant mean differences between devices, with limits of agreements comparable to those reported in similar comparison studies, nevertheless individual-level variability remained substantial. Conversely to the parameters mentioned before, RQ did differ significantly between devices. A stability analysis of the reference device showed that this discrepancy was most likely attributed to the sensor drift in the galvanic fuel cell oxygen sensor of the reference. An operating envelope analysis indicates that the prototype covers the ranges encountered in spontaneously breathing, mechanically ventilated and ECMO patients, and identifies the conditions that remain to be met in these configurations.
Together, these findings provide a foundation for breath-by-breath indirect calorimetry in clinical settings where it is currently restricted, and for more frequent metabolic measurements, including outside specialised clinical environments.
We are proud to announce that Taylor Frantz and his team at TARC (The Augmented Reality Company) have won the Jury Award at the MedTech Accelerator 2026, held in Brussels on 8 October.
TARC is developing an innovative augmented reality solution to support neurosurgeons in placing external ventricular drains (EVDs). By providing portable, image-guided navigation directly at the patient’s bedside, the technology aims to make neurosurgical interventions safer, faster, and more accurate.
The event also showcased other promising VUB health technology innovations. Ioana-Teodora Ouatu presented ORION, an innovative smart needle designed to improve safety and precision during IVF procedures. Colleagues from ELEC, Andy Keymolen and Antoine Marchal, under the guidance of John Lataire, participated with ROMULUS, a non-invasive technology for monitoring lung function.
The 2026 MedTech Accelerator was a successful edition for VUB HealthTech, highlighting the power of collaboration between engineers, researchers, and clinical teams in neurology, reproductive medicine, and pulmonology.
Congratulations to Taylor and the entire TARC team, and to all VUB participants!
More information:
Congratulations to Simon on this achievement!
FWOSB224: Runtime-Scalable 3D Representations through Spectrally Adaptive Primitives – Adrian Munteanu (Promotor)
Mihnea’s research focuses on making it faster and easier to turn photos and videos of the real world into realistic 3D scenes that can be explored on different devices. The goal is to develop a new way of representing these scenes that can automatically adapt their level of detail depending on the available computing power or internet bandwidth, without needing to rebuild them. Similar to how images and videos can be displayed at different resolutions, this approach would allow 3D content to be streamed and viewed at different quality levels.
On 26 October 2026 at 16:00, Alican ĂaÄlar will defend their PhD entitled âLOW-POWER CRYOGENIC CMOS INTERFACES FOR SCALABLE QUANTUM COMPUTERSâ.
Everybody is invited to attend the presentation in room D.2.01 or online via this link.
Realizing a fault-tolerant quantum computer based on superconducting or spin quantum bits (qubits) requires millions of physical qubits operating below 100 mK in dilution refrigerators. In the early years of quantum processor research and development, qubits were controlled using room temperature electronics due to their small scale. However, with the rapid increase in the number of qubits in the last decade, the massive amount of cabling required between mK temperatures and room temperature has become a bottleneck. To overcome this limitation, CMOS electronics, the technology underlying modern microprocessors, are increasingly used in close proximity to qubits at cryogenic temperatures, such as the mK and 4 K stages of dilution refrigerators. These circuits are commonly referred to as cryogenic CMOS (cryo-CMOS). The main challenge for scalable cryo-CMOS control electronics is the very low power budget in dilution refrigerators, due to their limited cooling power.
In the first phase of this research, a system-on-chip (SoC) for qubit readout at 4 K was developed using 28 nm bulk CMOS technology. The SoC employs a full zero intermediate frequency architecture and operates in the 6-8 GHz frequency range at 4 K. The readout SoC achieves lower power consumption with a noise performance comparable to prior works.
In the second phase, CMOS circuits operating at mK temperatures were investigated. In this context, two prototypes, a DC demultiplexer and a charge-redistribution DC digital-to-analog converter (DAC) implemented in 65 nm bulk CMOS technology, have been designed to alleviate the DC wiring bottleneck for spin qubit biasing. Thanks to their low power consumption, they enable the biasing of over 100,000 qubits at 100 mK.
Schuifelende Stapjes
Korte passen waarbij de voeten nauwelijks worden opgetild
Eenzijdige Klapvoet
U zet eerst de voorvoet op de grond in plaats van de hiel
Twijfel je nog? Je hoeft geen zelfdiagnose te stellen, laten we samen een videogesprek inplannen! Stuur ons een e-mail naar elena.botti@vub.be en wij kijken of je in aanmerking komt.
We willen loopanalyses eenvoudiger en toegankelijker maken. We vergelijken een kleine draagbare bewegingssensor (IMU) met een motion-capture-systeem voor laboratoriumgebruik (Vicon) om de methoden voor het analyseren van verschillende looppatronen te verbeteren. Op de onderstaande fotoâs ziet u de opstelling: de IMU is de kleine sensor op de onderrug van de patiĂ«nt en voor Vicon moeten er kleine markeringen op uw lichaam worden aangebracht.


đEĂ©n bezoek aan het Brubotics Rehabilitation Research Center (BRRC) â Laarbeeklaan 121, 1090 Jette, Brussel
â±ïžHet duurt ongeveer 2 uur vanaf het moment dat je binnenkomt tot het einde van de opnamesessie
đœDraag een sportbroekje of ondergoed, zodat we de markeringen kunnen aanbrengen. Je zal op blote voeten stappen
đ 15 wandelingen van 10 meter: 5 langzaam, 5 normaal en 5 snel, in je eigen tempo
âïžKlaar! Voor je het weet is alles afgerond, geen behandeling en geen vervolgafspraak. Je hebt bijgedragen aan het onderzoek!đŹ
Petits Pas TraĂźnants
Vous faites de petits pas et vous levez Ă peine les pieds du sol
Pied Tombant Unilatéral
Vous posez dâabord lâavant du pied au sol, plutĂŽt que le talon
Vous n’ĂȘtes pas sĂ»r(e)? Inutile de vous diagnostiquer vous-mĂȘme, organisons un appel vidĂ©o ensemble! Envoyez-nous un e-mail Ă l’adresse elena.botti@vub.be et nous vĂ©rifierons si vous ĂȘtes Ă©ligible.
Notre objectif est de rendre l’analyse de la marche plus simple et plus accessible. Pour se faire, nous voulons comparer un capteur de mouvement portable (IMU) Ă un systĂšme de laboratoire (Vicon) afin d’amĂ©liorer les mĂ©thodes d’analyse des diffĂ©rents types de marche en utilisant le systĂšme IMU. Le dispositif est visible sur les photos ci-dessous : l’IMU est le petit capteur placĂ© dans le bas du dos du patient, tandis que le systĂšme Vicon nĂ©cessite la pose de petits marqueurs sur votre corps.


đUne seule visite au Brubotics Rehabilitation Research Center (BRRC) â Avenue du Laerbeek 121, 1090 Jette, Bruxelles.
â±ïžEnviron 2 heures, depuis votre arrivĂ©e jusquâĂ la fin de la sĂ©ance dâenregistrement.
đœPortez un short de sport ou des sous-vĂȘtements afin que nous puissions placer les marqueurs. Vous marcherez pieds nus.
đ 15 allers-retours de 10 mĂštres : 5 lents, 5 normaux et 5 rapides, Ă votre rythme.
âïžC’est terminĂ© ! Vous aurez fini avant mĂȘme de vous en rendre compte, sans traitement ni suivi. Vous aurez contribuĂ© Ă la recherche !đŹ
On 23 October 2026 at 16:00, Brent De Weerdt will defend their PhD entitled âInterpretable Transformer Networks for Compressed Sensing, Representation Learning, and Distributed Source Codingâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
The current trend in artificial intelligence, and deep learning in particular, is to design increasingly large and complex models. Although these models achieve high accuracy, they require large datasets, high computational budgets, and often remain difficult to interpret. This thesis addresses these limitations by leveraging theoretical knowledge and data priors to design lightweight, performant, and inherently interpretable Transformer-based deep learning models.
The work first considers signal acquisition through compressed sensing. Using deep unfolding, the thesis designs neural networks for signal reconstruction from compressed measurements. By incorporating sparsity and temporal correlation, it develops interpretable Transformer models that reconstruct signals more accurately and efficiently than existing methods. The thesis then studies compact and privacy-preserving representations of human pose trajectories in video using contrastive learning and generative modelling, enabling anomaly detection through deviations from learned normal motion patterns. Finally, the thesis addresses distributed source coding, where correlated observations from multiple devices can be compressed more efficiently by exploiting inter-source dependencies. Building on LDPCA codes, it proposes a multi-rate Transformer decoder for Slepian-Wolf decoding that outperforms traditional decoders and reduces communication cost compared with alternative distributed compression methods.
Overall, the thesis develops deep learning architectures that incorporate signal processing, information theory, and data-specific prior knowledge, achieving state-of-the-art performance across several tasks while reducing the computational budget by at least an order of magnitude in most cases.
The planner you wish you had when facing a project deadline!
Nothing special. Anyone could make one â if only they had the time!
Need a quick project timeline without struggling with complicated planning software?
This simple Excel sheet lets you create and update project roadmaps on the fly. Just enter your tasks, dates and milestones, and the chart builds itself. Change something, and it updates automatically.
A few innocent VBA macros do the work behind the scenes. No installation, no complicated setup.
Download it, enable macros and start planning. Instructions are included in the Excel file.
On 30 October 2026 at 16:00, Silvia Zaccardi will defend their PhD entitled âBeyond the lab: Towards portable clinical motion analysis with observer-worn head-mounted displaysâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
Human movement analysis provides valuable information for assessing patients and monitoring rehabilitation. However, accurate measurements typically require expensive equipment in specialised laboratories, limiting their use in everyday clinical practice.
This thesis explores how Artificial Intelligence (AI) and Mixed Reality (MR) headsets could make movement analysis more affordable, portable, and accessible. The long-term vision is a system in which a physiotherapist wears an MR headset while a patient walks or performs exercises in front of them. Cameras built into the headset would track the patientâs movements, while relevant information, such as joint angles and walking characteristics, would be displayed directly in the physiotherapistâs field of view, supporting clinical assessment in real time.
Towards this vision, the research proposes two approaches to capturing human movement using the headsetâs built-in cameras and sensors. The first uses AI to estimate body movement directly from video, without requiring physical markers, while the second uses small reflective markers attached to the patientâs body to improve tracking accuracy. The thesis also explores AI algorithms to estimate the forces between the feet and the ground from movement data alone, potentially reducing the need for specialised force-measuring plates embedded in the laboratory floor.
The findings demonstrate the potential of these technologies while highlighting the trade-offs between measurement accuracy, ease of use, and real-time performance. Overall, this thesis contributes towards making objective movement analysis more accessible beyond specialised laboratories.
âSignal Processing in the AI eraâ was the tagline of this yearâs IEEE International Conference on Acoustics, Speech and Signal Processing, taking place in Rhodes, Greece.
In this context, Brent de Weerdt, Xiangyu Yang, Boris Joukovsky, Alex Stergiou and Nikos Deligiannis presented ETRO’s research during poster sessions and oral presentations, with novel ways to process and understand graph, video, and audio data. Nikos Deligiannis chaired a session on Graph Deep Learning, attended the IEEE T-IP Editorial Board Meeting, and had the opportunity to meet with collaborators from the VUB-Duke-Ugent-UCL joint lab.
Featured articles:

Lars Bonnefoy has been awarded an FWO PhD fellowship for his project âFrom Clouds to Kilowatts: Multimodal Deep Learning for Probabilistic Solar Irradiance Nowcasting and Solar Power Forecastingâ. With over 11 GW of solar capacity installed in Belgium, accurate forecasts for the next hours are increasingly important for grid stability. Lars will develop an open-source deep learning system combining satellite imagery, weather model output and ground measurements into probabilistic solar irradiance forecasts. The project is embedded in the Flanders Alliance for Climate Technology (FACT): together with the EVERGi group, the forecasts will be tested at the Smart Village Lab in Research Park Zellik. Lars is supervised by Prof. Lesley De Cruz (VUB-ETRO/RMI), with Prof. Thierry Coosemans (VUB-EVERGi) and Dr. Pierre de Buyl (RMI) as co-supervisors.
Congratulations, Lars!
A free trade agreement (FTA) is a framework. It is essential, but nobody lives in a structure. It still needs walls: institutions and rules companies can rely on. Windows: real market access, in both directions. Light: predictable, transparent conditions. Ventilation: talent and ideas moving freely. And movement: goods, investment and people actually flowing through the building.
That was the thread through the discussion. Beyond tariffs, the conversation kept returning to deeper partnerships in research, technology and innovation, and to the hard part: turning good research into solutions that can be tested, commercialised and scaled to benefit both sides.
The partnership works because each side brings what the other needs. India brings scale, talent and a fast-growing data ecosystem and an ability for rapid deployment. Belgium and the wider European Union bring research excellence and specialised industrial strengths, from life sciences and chemicals to clean technology. Belgium invests 3.4% of its gross domestic product in research and development, among the highest rates in Europe. Through Antwerp-Bruges, one of Europe’s largest ports, it offers a gateway to a single market of 450 million people, and European standards are trusted worldwide. Together, that creates real opportunities for companies on both sides, including small and medium-sized enterprises.
For India, the European Union is already one of its two largest trading partners. For Europe, India is still around 2% of its trade. That gap is the opportunity.
Thank you to Geoffroy Gueyffier , Asim Anwar, Erika Staël von Holstein, Renaat Schrooten , Edith N. NordmannTej Paul Bhatla, Prof. Dr. Johan Stiens ,Hemakiran Gupta , Jan Luykx and Frank Haak for conversations of real substance. And a special thank you to Rakesh Kapila and BICA, the Belgium India Creative Association, for bringing together such a diverse group of voices.
Ambassador Jan Luykx set the bar for success: if, in ten years, India and the European Union are shaping world geopolitics together, we will know the FTA has worked.
Research by Redona Brahimetaj and colleagues at ETRO-imec shows that the tissue surrounding breast microcalcifications can provide more diagnostic information than the calcium deposits themselves, opening new perspectives for breast cancer imaging.
Breast microcalcifications are tiny calcium deposits that can appear in breast tissue and are important indicators in breast cancer screening. Traditionally, imaging techniques have focused on the shape, size and distribution of these deposits to help distinguish benign from malignant lesions.
A new study led by Redona Brahimetaj at the Department of Electronics and Informatics (ETRO-imec) of Vrije Universiteit Brussel challenges this perspective. The research demonstrates that the tissue immediately surrounding individual microcalcifications contains valuable diagnostic information, potentially even more than the calcifications themselves.
Using high-resolution three-dimensional micro-computed tomography (micro-CT), the research team analysed 3,504 individual microcalcifications from 94 breast biopsy specimens, achieving a spatial resolution of approximately 8 micrometres.
By combining advanced medical image analysis, radiomics and machine-learning techniques, the researchers systematically investigated where the most useful diagnostic information originates: within the calcium deposits or in the surrounding breast tissue.
The results were striking. Including the surrounding tissue improved classification performance, with the area under the receiver operating characteristic curve (AUC) increasing from approximately 0.69 to 0.81.
Remarkably, analysing the surrounding tissue alone achieved comparable performance, demonstrating that the microenvironment around the calcifications is a particularly important source of diagnostic information.
This work highlights ETRO’s expertise in medical image processing, quantitative image analysis and artificial intelligence, and illustrates how advanced computational methods can reveal information that conventional imaging analysis may overlook.
By shifting attention from the calcifications themselves to the interface between calcium deposits and surrounding tissue, the research provides a new direction for developing quantitative imaging biomarkers.
Although the study used highly detailed images of biopsy specimens rather than conventional clinical breast images, its findings offer important insights that may guide future breast cancer imaging research. Further validation and investigation of the underlying tissue structures will be necessary before clinical applications can be considered.
The study was conducted through a collaboration between ETRO-imec, VUB’s Laboratory of Experimental Pathology, and the Departments of Pathology and Radiology at UZ Brussel.
The research brings together expertise in artificial intelligence, medical imaging, radiology and pathology, demonstrating the value of interdisciplinary collaboration in addressing challenging medical diagnostic problems.
Micro-CT Radiomics Reveals the Microcalcification-Surrounding Tissue Interface as the Diagnostic Epicenter of Breast Microcalcifications
Authors: Redona Brahimetaj, Jan Cornelis, Guillaume Courtoy, Wim Waelput, Inneke Willekens, Johan De Mey, Ramses Forsyth and Bart Jansen.
https://www.vub.be/en/news/beyond-calcification-vub-imec-researchers-identify
On 30 October 2026 at 16:00, Sebastiaan Priem will defend their PhD entitled âBREATH-BY-BREATH INDIRECT CALORIMETRY BASED ON A DYNAMIC HALDANE TRANSFORMATION: FROM THEORETICAL FRAMEWORK TO CLINICALLY EVALUATED PROTOTYPEâ.
Everybody is invited to attend the presentation in room K.2.56 or online via this link.
Abstract â
Indirect calorimetry is considered the gold standard for measuring resting energy expenditure and respiratory quotient and relies on the measurement of a patient’s rate of oxygen consumption and carbon dioxide production. Despite its established value, its use is constrained by several limitations. First, the use of the Haldane transformation to infer inspired flow, amplifies measurement errors as the inspired oxygen fraction increases, which prevents measurements in patients receiving more than 60-70 % oxygen. Secondly, a single short-term measurement is typically extrapolated to represent the energy expenditure of the patient over a period of 24 hours, which doesn’t capture intra- and inter-day variability. Finally, the cost, form factor, maintenance burden and requirement of trained personnel of available devices confine their use to specialised clinical or research environments.
This dissertation establishes a theoretical framework for breath-by-breath indirect calorimetry that determines the oxygen consumption rate without measuring the expired oxygen fraction and without assuming a value for the respiratory quotient. When both the inspired and expired flow rates are measured, the expired inert gas fraction follows from the mass balance of inert gases, and the expired oxygen fraction from the composition of the expired gas. By applying this method, the inspired oxygen fraction no longer appears in a denominator, which analytically avoids the error amplification at high inspired oxygen fractions. Additionally, a dynamic Haldane transformation is introduced, which distributes the volume correction over the exhalation according to the expired CO2 waveform, and thereby accounts for the dead-space gas at the start of each exhalation.
The requirements for breath-by-breath gas exchange measurements in the clinic and at home are derived from a criterion on the error of REE, and the sensing principles for each quantity are evaluated against them. This analysis shows that flow and CO2 can be measured directly, whereas the expired oxygen fraction cannot, and leads to a compact prototype that measures flow and CO2 at the airway opening.
In a pilot comparison with a clinical reference calorimeter (Q-NRG, COSMED) in 12 healthy volunteers, the agreement for REE and gas exchange rates showed no statistically significant mean differences between devices, with limits of agreements comparable to those reported in similar comparison studies, nevertheless individual-level variability remained substantial. Conversely to the parameters mentioned before, RQ did differ significantly between devices. A stability analysis of the reference device showed that this discrepancy was most likely attributed to the sensor drift in the galvanic fuel cell oxygen sensor of the reference. An operating envelope analysis indicates that the prototype covers the ranges encountered in spontaneously breathing, mechanically ventilated and ECMO patients, and identifies the conditions that remain to be met in these configurations.
Together, these findings provide a foundation for breath-by-breath indirect calorimetry in clinical settings where it is currently restricted, and for more frequent metabolic measurements, including outside specialised clinical environments.
We are proud to announce that Taylor Frantz and his team at TARC (The Augmented Reality Company) have won the Jury Award at the MedTech Accelerator 2026, held in Brussels on 8 October.
TARC is developing an innovative augmented reality solution to support neurosurgeons in placing external ventricular drains (EVDs). By providing portable, image-guided navigation directly at the patient’s bedside, the technology aims to make neurosurgical interventions safer, faster, and more accurate.
The event also showcased other promising VUB health technology innovations. Ioana-Teodora Ouatu presented ORION, an innovative smart needle designed to improve safety and precision during IVF procedures. Colleagues from ELEC, Andy Keymolen and Antoine Marchal, under the guidance of John Lataire, participated with ROMULUS, a non-invasive technology for monitoring lung function.
The 2026 MedTech Accelerator was a successful edition for VUB HealthTech, highlighting the power of collaboration between engineers, researchers, and clinical teams in neurology, reproductive medicine, and pulmonology.
Congratulations to Taylor and the entire TARC team, and to all VUB participants!
More information:
Congratulations to Simon on this achievement!
FWOSB224: Runtime-Scalable 3D Representations through Spectrally Adaptive Primitives – Adrian Munteanu (Promotor)
Mihnea’s research focuses on making it faster and easier to turn photos and videos of the real world into realistic 3D scenes that can be explored on different devices. The goal is to develop a new way of representing these scenes that can automatically adapt their level of detail depending on the available computing power or internet bandwidth, without needing to rebuild them. Similar to how images and videos can be displayed at different resolutions, this approach would allow 3D content to be streamed and viewed at different quality levels.
On 26 October 2026 at 16:00, Alican ĂaÄlar will defend their PhD entitled âLOW-POWER CRYOGENIC CMOS INTERFACES FOR SCALABLE QUANTUM COMPUTERSâ.
Everybody is invited to attend the presentation in room D.2.01 or online via this link.
Realizing a fault-tolerant quantum computer based on superconducting or spin quantum bits (qubits) requires millions of physical qubits operating below 100 mK in dilution refrigerators. In the early years of quantum processor research and development, qubits were controlled using room temperature electronics due to their small scale. However, with the rapid increase in the number of qubits in the last decade, the massive amount of cabling required between mK temperatures and room temperature has become a bottleneck. To overcome this limitation, CMOS electronics, the technology underlying modern microprocessors, are increasingly used in close proximity to qubits at cryogenic temperatures, such as the mK and 4 K stages of dilution refrigerators. These circuits are commonly referred to as cryogenic CMOS (cryo-CMOS). The main challenge for scalable cryo-CMOS control electronics is the very low power budget in dilution refrigerators, due to their limited cooling power.
In the first phase of this research, a system-on-chip (SoC) for qubit readout at 4 K was developed using 28 nm bulk CMOS technology. The SoC employs a full zero intermediate frequency architecture and operates in the 6-8 GHz frequency range at 4 K. The readout SoC achieves lower power consumption with a noise performance comparable to prior works.
In the second phase, CMOS circuits operating at mK temperatures were investigated. In this context, two prototypes, a DC demultiplexer and a charge-redistribution DC digital-to-analog converter (DAC) implemented in 65 nm bulk CMOS technology, have been designed to alleviate the DC wiring bottleneck for spin qubit biasing. Thanks to their low power consumption, they enable the biasing of over 100,000 qubits at 100 mK.
Schuifelende Stapjes
Korte passen waarbij de voeten nauwelijks worden opgetild
Eenzijdige Klapvoet
U zet eerst de voorvoet op de grond in plaats van de hiel
Twijfel je nog? Je hoeft geen zelfdiagnose te stellen, laten we samen een videogesprek inplannen! Stuur ons een e-mail naar elena.botti@vub.be en wij kijken of je in aanmerking komt.
We willen loopanalyses eenvoudiger en toegankelijker maken. We vergelijken een kleine draagbare bewegingssensor (IMU) met een motion-capture-systeem voor laboratoriumgebruik (Vicon) om de methoden voor het analyseren van verschillende looppatronen te verbeteren. Op de onderstaande fotoâs ziet u de opstelling: de IMU is de kleine sensor op de onderrug van de patiĂ«nt en voor Vicon moeten er kleine markeringen op uw lichaam worden aangebracht.


đEĂ©n bezoek aan het Brubotics Rehabilitation Research Center (BRRC) â Laarbeeklaan 121, 1090 Jette, Brussel
â±ïžHet duurt ongeveer 2 uur vanaf het moment dat je binnenkomt tot het einde van de opnamesessie
đœDraag een sportbroekje of ondergoed, zodat we de markeringen kunnen aanbrengen. Je zal op blote voeten stappen
đ 15 wandelingen van 10 meter: 5 langzaam, 5 normaal en 5 snel, in je eigen tempo
âïžKlaar! Voor je het weet is alles afgerond, geen behandeling en geen vervolgafspraak. Je hebt bijgedragen aan het onderzoek!đŹ
Petits Pas TraĂźnants
Vous faites de petits pas et vous levez Ă peine les pieds du sol
Pied Tombant Unilatéral
Vous posez dâabord lâavant du pied au sol, plutĂŽt que le talon
Vous n’ĂȘtes pas sĂ»r(e)? Inutile de vous diagnostiquer vous-mĂȘme, organisons un appel vidĂ©o ensemble! Envoyez-nous un e-mail Ă l’adresse elena.botti@vub.be et nous vĂ©rifierons si vous ĂȘtes Ă©ligible.
Notre objectif est de rendre l’analyse de la marche plus simple et plus accessible. Pour se faire, nous voulons comparer un capteur de mouvement portable (IMU) Ă un systĂšme de laboratoire (Vicon) afin d’amĂ©liorer les mĂ©thodes d’analyse des diffĂ©rents types de marche en utilisant le systĂšme IMU. Le dispositif est visible sur les photos ci-dessous : l’IMU est le petit capteur placĂ© dans le bas du dos du patient, tandis que le systĂšme Vicon nĂ©cessite la pose de petits marqueurs sur votre corps.


đUne seule visite au Brubotics Rehabilitation Research Center (BRRC) â Avenue du Laerbeek 121, 1090 Jette, Bruxelles.
â±ïžEnviron 2 heures, depuis votre arrivĂ©e jusquâĂ la fin de la sĂ©ance dâenregistrement.
đœPortez un short de sport ou des sous-vĂȘtements afin que nous puissions placer les marqueurs. Vous marcherez pieds nus.
đ 15 allers-retours de 10 mĂštres : 5 lents, 5 normaux et 5 rapides, Ă votre rythme.
âïžC’est terminĂ© ! Vous aurez fini avant mĂȘme de vous en rendre compte, sans traitement ni suivi. Vous aurez contribuĂ© Ă la recherche !đŹ
On 23 October 2026 at 16:00, Brent De Weerdt will defend their PhD entitled âInterpretable Transformer Networks for Compressed Sensing, Representation Learning, and Distributed Source Codingâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
The current trend in artificial intelligence, and deep learning in particular, is to design increasingly large and complex models. Although these models achieve high accuracy, they require large datasets, high computational budgets, and often remain difficult to interpret. This thesis addresses these limitations by leveraging theoretical knowledge and data priors to design lightweight, performant, and inherently interpretable Transformer-based deep learning models.
The work first considers signal acquisition through compressed sensing. Using deep unfolding, the thesis designs neural networks for signal reconstruction from compressed measurements. By incorporating sparsity and temporal correlation, it develops interpretable Transformer models that reconstruct signals more accurately and efficiently than existing methods. The thesis then studies compact and privacy-preserving representations of human pose trajectories in video using contrastive learning and generative modelling, enabling anomaly detection through deviations from learned normal motion patterns. Finally, the thesis addresses distributed source coding, where correlated observations from multiple devices can be compressed more efficiently by exploiting inter-source dependencies. Building on LDPCA codes, it proposes a multi-rate Transformer decoder for Slepian-Wolf decoding that outperforms traditional decoders and reduces communication cost compared with alternative distributed compression methods.
Overall, the thesis develops deep learning architectures that incorporate signal processing, information theory, and data-specific prior knowledge, achieving state-of-the-art performance across several tasks while reducing the computational budget by at least an order of magnitude in most cases.
The planner you wish you had when facing a project deadline!
Nothing special. Anyone could make one â if only they had the time!
Need a quick project timeline without struggling with complicated planning software?
This simple Excel sheet lets you create and update project roadmaps on the fly. Just enter your tasks, dates and milestones, and the chart builds itself. Change something, and it updates automatically.
A few innocent VBA macros do the work behind the scenes. No installation, no complicated setup.
Download it, enable macros and start planning. Instructions are included in the Excel file.
On 30 October 2026 at 16:00, Silvia Zaccardi will defend their PhD entitled âBeyond the lab: Towards portable clinical motion analysis with observer-worn head-mounted displaysâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
Human movement analysis provides valuable information for assessing patients and monitoring rehabilitation. However, accurate measurements typically require expensive equipment in specialised laboratories, limiting their use in everyday clinical practice.
This thesis explores how Artificial Intelligence (AI) and Mixed Reality (MR) headsets could make movement analysis more affordable, portable, and accessible. The long-term vision is a system in which a physiotherapist wears an MR headset while a patient walks or performs exercises in front of them. Cameras built into the headset would track the patientâs movements, while relevant information, such as joint angles and walking characteristics, would be displayed directly in the physiotherapistâs field of view, supporting clinical assessment in real time.
Towards this vision, the research proposes two approaches to capturing human movement using the headsetâs built-in cameras and sensors. The first uses AI to estimate body movement directly from video, without requiring physical markers, while the second uses small reflective markers attached to the patientâs body to improve tracking accuracy. The thesis also explores AI algorithms to estimate the forces between the feet and the ground from movement data alone, potentially reducing the need for specialised force-measuring plates embedded in the laboratory floor.
The findings demonstrate the potential of these technologies while highlighting the trade-offs between measurement accuracy, ease of use, and real-time performance. Overall, this thesis contributes towards making objective movement analysis more accessible beyond specialised laboratories.
âSignal Processing in the AI eraâ was the tagline of this yearâs IEEE International Conference on Acoustics, Speech and Signal Processing, taking place in Rhodes, Greece.
In this context, Brent de Weerdt, Xiangyu Yang, Boris Joukovsky, Alex Stergiou and Nikos Deligiannis presented ETRO’s research during poster sessions and oral presentations, with novel ways to process and understand graph, video, and audio data. Nikos Deligiannis chaired a session on Graph Deep Learning, attended the IEEE T-IP Editorial Board Meeting, and had the opportunity to meet with collaborators from the VUB-Duke-Ugent-UCL joint lab.
Featured articles:

Lars Bonnefoy has been awarded an FWO PhD fellowship for his project âFrom Clouds to Kilowatts: Multimodal Deep Learning for Probabilistic Solar Irradiance Nowcasting and Solar Power Forecastingâ. With over 11 GW of solar capacity installed in Belgium, accurate forecasts for the next hours are increasingly important for grid stability. Lars will develop an open-source deep learning system combining satellite imagery, weather model output and ground measurements into probabilistic solar irradiance forecasts. The project is embedded in the Flanders Alliance for Climate Technology (FACT): together with the EVERGi group, the forecasts will be tested at the Smart Village Lab in Research Park Zellik. Lars is supervised by Prof. Lesley De Cruz (VUB-ETRO/RMI), with Prof. Thierry Coosemans (VUB-EVERGi) and Dr. Pierre de Buyl (RMI) as co-supervisors.
Congratulations, Lars!
A free trade agreement (FTA) is a framework. It is essential, but nobody lives in a structure. It still needs walls: institutions and rules companies can rely on. Windows: real market access, in both directions. Light: predictable, transparent conditions. Ventilation: talent and ideas moving freely. And movement: goods, investment and people actually flowing through the building.
That was the thread through the discussion. Beyond tariffs, the conversation kept returning to deeper partnerships in research, technology and innovation, and to the hard part: turning good research into solutions that can be tested, commercialised and scaled to benefit both sides.
The partnership works because each side brings what the other needs. India brings scale, talent and a fast-growing data ecosystem and an ability for rapid deployment. Belgium and the wider European Union bring research excellence and specialised industrial strengths, from life sciences and chemicals to clean technology. Belgium invests 3.4% of its gross domestic product in research and development, among the highest rates in Europe. Through Antwerp-Bruges, one of Europe’s largest ports, it offers a gateway to a single market of 450 million people, and European standards are trusted worldwide. Together, that creates real opportunities for companies on both sides, including small and medium-sized enterprises.
For India, the European Union is already one of its two largest trading partners. For Europe, India is still around 2% of its trade. That gap is the opportunity.
Thank you to Geoffroy Gueyffier , Asim Anwar, Erika Staël von Holstein, Renaat Schrooten , Edith N. NordmannTej Paul Bhatla, Prof. Dr. Johan Stiens ,Hemakiran Gupta , Jan Luykx and Frank Haak for conversations of real substance. And a special thank you to Rakesh Kapila and BICA, the Belgium India Creative Association, for bringing together such a diverse group of voices.
Ambassador Jan Luykx set the bar for success: if, in ten years, India and the European Union are shaping world geopolitics together, we will know the FTA has worked.
Research by Redona Brahimetaj and colleagues at ETRO-imec shows that the tissue surrounding breast microcalcifications can provide more diagnostic information than the calcium deposits themselves, opening new perspectives for breast cancer imaging.
Breast microcalcifications are tiny calcium deposits that can appear in breast tissue and are important indicators in breast cancer screening. Traditionally, imaging techniques have focused on the shape, size and distribution of these deposits to help distinguish benign from malignant lesions.
A new study led by Redona Brahimetaj at the Department of Electronics and Informatics (ETRO-imec) of Vrije Universiteit Brussel challenges this perspective. The research demonstrates that the tissue immediately surrounding individual microcalcifications contains valuable diagnostic information, potentially even more than the calcifications themselves.
Using high-resolution three-dimensional micro-computed tomography (micro-CT), the research team analysed 3,504 individual microcalcifications from 94 breast biopsy specimens, achieving a spatial resolution of approximately 8 micrometres.
By combining advanced medical image analysis, radiomics and machine-learning techniques, the researchers systematically investigated where the most useful diagnostic information originates: within the calcium deposits or in the surrounding breast tissue.
The results were striking. Including the surrounding tissue improved classification performance, with the area under the receiver operating characteristic curve (AUC) increasing from approximately 0.69 to 0.81.
Remarkably, analysing the surrounding tissue alone achieved comparable performance, demonstrating that the microenvironment around the calcifications is a particularly important source of diagnostic information.
This work highlights ETRO’s expertise in medical image processing, quantitative image analysis and artificial intelligence, and illustrates how advanced computational methods can reveal information that conventional imaging analysis may overlook.
By shifting attention from the calcifications themselves to the interface between calcium deposits and surrounding tissue, the research provides a new direction for developing quantitative imaging biomarkers.
Although the study used highly detailed images of biopsy specimens rather than conventional clinical breast images, its findings offer important insights that may guide future breast cancer imaging research. Further validation and investigation of the underlying tissue structures will be necessary before clinical applications can be considered.
The study was conducted through a collaboration between ETRO-imec, VUB’s Laboratory of Experimental Pathology, and the Departments of Pathology and Radiology at UZ Brussel.
The research brings together expertise in artificial intelligence, medical imaging, radiology and pathology, demonstrating the value of interdisciplinary collaboration in addressing challenging medical diagnostic problems.
Micro-CT Radiomics Reveals the Microcalcification-Surrounding Tissue Interface as the Diagnostic Epicenter of Breast Microcalcifications
Authors: Redona Brahimetaj, Jan Cornelis, Guillaume Courtoy, Wim Waelput, Inneke Willekens, Johan De Mey, Ramses Forsyth and Bart Jansen.
https://www.vub.be/en/news/beyond-calcification-vub-imec-researchers-identify
On 30 October 2026 at 16:00, Sebastiaan Priem will defend their PhD entitled âBREATH-BY-BREATH INDIRECT CALORIMETRY BASED ON A DYNAMIC HALDANE TRANSFORMATION: FROM THEORETICAL FRAMEWORK TO CLINICALLY EVALUATED PROTOTYPEâ.
Everybody is invited to attend the presentation in room K.2.56 or online via this link.
Abstract â
Indirect calorimetry is considered the gold standard for measuring resting energy expenditure and respiratory quotient and relies on the measurement of a patient’s rate of oxygen consumption and carbon dioxide production. Despite its established value, its use is constrained by several limitations. First, the use of the Haldane transformation to infer inspired flow, amplifies measurement errors as the inspired oxygen fraction increases, which prevents measurements in patients receiving more than 60-70 % oxygen. Secondly, a single short-term measurement is typically extrapolated to represent the energy expenditure of the patient over a period of 24 hours, which doesn’t capture intra- and inter-day variability. Finally, the cost, form factor, maintenance burden and requirement of trained personnel of available devices confine their use to specialised clinical or research environments.
This dissertation establishes a theoretical framework for breath-by-breath indirect calorimetry that determines the oxygen consumption rate without measuring the expired oxygen fraction and without assuming a value for the respiratory quotient. When both the inspired and expired flow rates are measured, the expired inert gas fraction follows from the mass balance of inert gases, and the expired oxygen fraction from the composition of the expired gas. By applying this method, the inspired oxygen fraction no longer appears in a denominator, which analytically avoids the error amplification at high inspired oxygen fractions. Additionally, a dynamic Haldane transformation is introduced, which distributes the volume correction over the exhalation according to the expired CO2 waveform, and thereby accounts for the dead-space gas at the start of each exhalation.
The requirements for breath-by-breath gas exchange measurements in the clinic and at home are derived from a criterion on the error of REE, and the sensing principles for each quantity are evaluated against them. This analysis shows that flow and CO2 can be measured directly, whereas the expired oxygen fraction cannot, and leads to a compact prototype that measures flow and CO2 at the airway opening.
In a pilot comparison with a clinical reference calorimeter (Q-NRG, COSMED) in 12 healthy volunteers, the agreement for REE and gas exchange rates showed no statistically significant mean differences between devices, with limits of agreements comparable to those reported in similar comparison studies, nevertheless individual-level variability remained substantial. Conversely to the parameters mentioned before, RQ did differ significantly between devices. A stability analysis of the reference device showed that this discrepancy was most likely attributed to the sensor drift in the galvanic fuel cell oxygen sensor of the reference. An operating envelope analysis indicates that the prototype covers the ranges encountered in spontaneously breathing, mechanically ventilated and ECMO patients, and identifies the conditions that remain to be met in these configurations.
Together, these findings provide a foundation for breath-by-breath indirect calorimetry in clinical settings where it is currently restricted, and for more frequent metabolic measurements, including outside specialised clinical environments.
We are proud to announce that Taylor Frantz and his team at TARC (The Augmented Reality Company) have won the Jury Award at the MedTech Accelerator 2026, held in Brussels on 8 October.
TARC is developing an innovative augmented reality solution to support neurosurgeons in placing external ventricular drains (EVDs). By providing portable, image-guided navigation directly at the patient’s bedside, the technology aims to make neurosurgical interventions safer, faster, and more accurate.
The event also showcased other promising VUB health technology innovations. Ioana-Teodora Ouatu presented ORION, an innovative smart needle designed to improve safety and precision during IVF procedures. Colleagues from ELEC, Andy Keymolen and Antoine Marchal, under the guidance of John Lataire, participated with ROMULUS, a non-invasive technology for monitoring lung function.
The 2026 MedTech Accelerator was a successful edition for VUB HealthTech, highlighting the power of collaboration between engineers, researchers, and clinical teams in neurology, reproductive medicine, and pulmonology.
Congratulations to Taylor and the entire TARC team, and to all VUB participants!
More information:
Congratulations to Simon on this achievement!
FWOSB224: Runtime-Scalable 3D Representations through Spectrally Adaptive Primitives – Adrian Munteanu (Promotor)
Mihnea’s research focuses on making it faster and easier to turn photos and videos of the real world into realistic 3D scenes that can be explored on different devices. The goal is to develop a new way of representing these scenes that can automatically adapt their level of detail depending on the available computing power or internet bandwidth, without needing to rebuild them. Similar to how images and videos can be displayed at different resolutions, this approach would allow 3D content to be streamed and viewed at different quality levels.
On 26 October 2026 at 16:00, Alican ĂaÄlar will defend their PhD entitled âLOW-POWER CRYOGENIC CMOS INTERFACES FOR SCALABLE QUANTUM COMPUTERSâ.
Everybody is invited to attend the presentation in room D.2.01 or online via this link.
Realizing a fault-tolerant quantum computer based on superconducting or spin quantum bits (qubits) requires millions of physical qubits operating below 100 mK in dilution refrigerators. In the early years of quantum processor research and development, qubits were controlled using room temperature electronics due to their small scale. However, with the rapid increase in the number of qubits in the last decade, the massive amount of cabling required between mK temperatures and room temperature has become a bottleneck. To overcome this limitation, CMOS electronics, the technology underlying modern microprocessors, are increasingly used in close proximity to qubits at cryogenic temperatures, such as the mK and 4 K stages of dilution refrigerators. These circuits are commonly referred to as cryogenic CMOS (cryo-CMOS). The main challenge for scalable cryo-CMOS control electronics is the very low power budget in dilution refrigerators, due to their limited cooling power.
In the first phase of this research, a system-on-chip (SoC) for qubit readout at 4 K was developed using 28 nm bulk CMOS technology. The SoC employs a full zero intermediate frequency architecture and operates in the 6-8 GHz frequency range at 4 K. The readout SoC achieves lower power consumption with a noise performance comparable to prior works.
In the second phase, CMOS circuits operating at mK temperatures were investigated. In this context, two prototypes, a DC demultiplexer and a charge-redistribution DC digital-to-analog converter (DAC) implemented in 65 nm bulk CMOS technology, have been designed to alleviate the DC wiring bottleneck for spin qubit biasing. Thanks to their low power consumption, they enable the biasing of over 100,000 qubits at 100 mK.
Schuifelende Stapjes
Korte passen waarbij de voeten nauwelijks worden opgetild
Eenzijdige Klapvoet
U zet eerst de voorvoet op de grond in plaats van de hiel
Twijfel je nog? Je hoeft geen zelfdiagnose te stellen, laten we samen een videogesprek inplannen! Stuur ons een e-mail naar elena.botti@vub.be en wij kijken of je in aanmerking komt.
We willen loopanalyses eenvoudiger en toegankelijker maken. We vergelijken een kleine draagbare bewegingssensor (IMU) met een motion-capture-systeem voor laboratoriumgebruik (Vicon) om de methoden voor het analyseren van verschillende looppatronen te verbeteren. Op de onderstaande fotoâs ziet u de opstelling: de IMU is de kleine sensor op de onderrug van de patiĂ«nt en voor Vicon moeten er kleine markeringen op uw lichaam worden aangebracht.


đEĂ©n bezoek aan het Brubotics Rehabilitation Research Center (BRRC) â Laarbeeklaan 121, 1090 Jette, Brussel
â±ïžHet duurt ongeveer 2 uur vanaf het moment dat je binnenkomt tot het einde van de opnamesessie
đœDraag een sportbroekje of ondergoed, zodat we de markeringen kunnen aanbrengen. Je zal op blote voeten stappen
đ 15 wandelingen van 10 meter: 5 langzaam, 5 normaal en 5 snel, in je eigen tempo
âïžKlaar! Voor je het weet is alles afgerond, geen behandeling en geen vervolgafspraak. Je hebt bijgedragen aan het onderzoek!đŹ
Petits Pas TraĂźnants
Vous faites de petits pas et vous levez Ă peine les pieds du sol
Pied Tombant Unilatéral
Vous posez dâabord lâavant du pied au sol, plutĂŽt que le talon
Vous n’ĂȘtes pas sĂ»r(e)? Inutile de vous diagnostiquer vous-mĂȘme, organisons un appel vidĂ©o ensemble! Envoyez-nous un e-mail Ă l’adresse elena.botti@vub.be et nous vĂ©rifierons si vous ĂȘtes Ă©ligible.
Notre objectif est de rendre l’analyse de la marche plus simple et plus accessible. Pour se faire, nous voulons comparer un capteur de mouvement portable (IMU) Ă un systĂšme de laboratoire (Vicon) afin d’amĂ©liorer les mĂ©thodes d’analyse des diffĂ©rents types de marche en utilisant le systĂšme IMU. Le dispositif est visible sur les photos ci-dessous : l’IMU est le petit capteur placĂ© dans le bas du dos du patient, tandis que le systĂšme Vicon nĂ©cessite la pose de petits marqueurs sur votre corps.


đUne seule visite au Brubotics Rehabilitation Research Center (BRRC) â Avenue du Laerbeek 121, 1090 Jette, Bruxelles.
â±ïžEnviron 2 heures, depuis votre arrivĂ©e jusquâĂ la fin de la sĂ©ance dâenregistrement.
đœPortez un short de sport ou des sous-vĂȘtements afin que nous puissions placer les marqueurs. Vous marcherez pieds nus.
đ 15 allers-retours de 10 mĂštres : 5 lents, 5 normaux et 5 rapides, Ă votre rythme.
âïžC’est terminĂ© ! Vous aurez fini avant mĂȘme de vous en rendre compte, sans traitement ni suivi. Vous aurez contribuĂ© Ă la recherche !đŹ
On 23 October 2026 at 16:00, Brent De Weerdt will defend their PhD entitled âInterpretable Transformer Networks for Compressed Sensing, Representation Learning, and Distributed Source Codingâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
The current trend in artificial intelligence, and deep learning in particular, is to design increasingly large and complex models. Although these models achieve high accuracy, they require large datasets, high computational budgets, and often remain difficult to interpret. This thesis addresses these limitations by leveraging theoretical knowledge and data priors to design lightweight, performant, and inherently interpretable Transformer-based deep learning models.
The work first considers signal acquisition through compressed sensing. Using deep unfolding, the thesis designs neural networks for signal reconstruction from compressed measurements. By incorporating sparsity and temporal correlation, it develops interpretable Transformer models that reconstruct signals more accurately and efficiently than existing methods. The thesis then studies compact and privacy-preserving representations of human pose trajectories in video using contrastive learning and generative modelling, enabling anomaly detection through deviations from learned normal motion patterns. Finally, the thesis addresses distributed source coding, where correlated observations from multiple devices can be compressed more efficiently by exploiting inter-source dependencies. Building on LDPCA codes, it proposes a multi-rate Transformer decoder for Slepian-Wolf decoding that outperforms traditional decoders and reduces communication cost compared with alternative distributed compression methods.
Overall, the thesis develops deep learning architectures that incorporate signal processing, information theory, and data-specific prior knowledge, achieving state-of-the-art performance across several tasks while reducing the computational budget by at least an order of magnitude in most cases.
The planner you wish you had when facing a project deadline!
Nothing special. Anyone could make one â if only they had the time!
Need a quick project timeline without struggling with complicated planning software?
This simple Excel sheet lets you create and update project roadmaps on the fly. Just enter your tasks, dates and milestones, and the chart builds itself. Change something, and it updates automatically.
A few innocent VBA macros do the work behind the scenes. No installation, no complicated setup.
Download it, enable macros and start planning. Instructions are included in the Excel file.
On 30 October 2026 at 16:00, Silvia Zaccardi will defend their PhD entitled âBeyond the lab: Towards portable clinical motion analysis with observer-worn head-mounted displaysâ.
Everybody is invited to attend the presentation in room LIC.0.04 or online via this link.
Human movement analysis provides valuable information for assessing patients and monitoring rehabilitation. However, accurate measurements typically require expensive equipment in specialised laboratories, limiting their use in everyday clinical practice.
This thesis explores how Artificial Intelligence (AI) and Mixed Reality (MR) headsets could make movement analysis more affordable, portable, and accessible. The long-term vision is a system in which a physiotherapist wears an MR headset while a patient walks or performs exercises in front of them. Cameras built into the headset would track the patientâs movements, while relevant information, such as joint angles and walking characteristics, would be displayed directly in the physiotherapistâs field of view, supporting clinical assessment in real time.
Towards this vision, the research proposes two approaches to capturing human movement using the headsetâs built-in cameras and sensors. The first uses AI to estimate body movement directly from video, without requiring physical markers, while the second uses small reflective markers attached to the patientâs body to improve tracking accuracy. The thesis also explores AI algorithms to estimate the forces between the feet and the ground from movement data alone, potentially reducing the need for specialised force-measuring plates embedded in the laboratory floor.
The findings demonstrate the potential of these technologies while highlighting the trade-offs between measurement accuracy, ease of use, and real-time performance. Overall, this thesis contributes towards making objective movement analysis more accessible beyond specialised laboratories.