Postdoctoral researcher at University of Geneva (80%) & University Hospital of Bern (20%)
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
I am a postdoctoral researcher in UNIGE. I hold a PhD in Electrical Engineering (Institute of Bioengineering) from École Polytechnique Fédérale de Lausanne (EPFL). I am motivated and passionate about brain research (biomedical engineering & computational neuroscience). My research focuses on developing appropriate statistical frameworks for the investigation of the functional and structural brain connectivity alterations in preterm newborns.
June 2021 - November 2021
October 2017 - May 2021
- PhD Student
- PhD student & teaching assistant (in the context of my PhD studies) for Image Processing I, Image Processing II and Signal processing for functional brain imaging courses.
Neonatal Intensive Care Units (NICU) provide special equipment designed to give life support for the increasing number of prematurely born infants and assure their survival. More recently NICU's strive to include developmentally oriented care and modulate sensory input for preterm infants. Music, among other sensory stimuli, has been introduced int...
Premature birth has been associated with poor neurodevelopmental outcomes. However, the relation between such outcomes and brain growth in the neonatal period has not yet been fully elucidated. This study investigates longitudinal brain development between birth and term-equivalent age (TEA) by quantitative imaging in a cohort of premature infants...
Music is known to induce emotions and activate associated memories, including musical memories. In adults, it is well known that music activates both working memory and limbic networks. We have recently discovered that as early as during the newborn period, familiar music is processed differently from unfamiliar music. The present study evaluates m...
Prematurity disrupts brain development during a critical period of brain growth and organization and is known to be associated with an increased risk of neurodevelopmental impairments. Investigating whole-brain structural connectivity alterations accompanying preterm birth may provide a better comprehension of the neurobiological mechanisms related...
In this study we explore the effects of an early intervention on functional connectivity in preterm newborns. A group of preterm newborns underwent musical intervention in NICU during hospitalization. At TEA, two resting-state fMRI runs were collected. The first was before and the second after the presentation of the same musical stimulus used for...
In this study, we combined complex network theory with machine learning in order to grasp potential biomarkers of brain development. The data consists of brain connectomes (brain connectivity matrices) of 53 children aged six years old. For each subject, we estimated brain network-based measures at four different levels: connection, node, module an...
Neural and functional consequences of prematurity warrant for consistent neonatal intensive care unit(NICU) enhancement. Appropriate environmental enrichment has been shown to have a positive impact on behavioral states and brain development. Over the past decade, music in NICU has been shown to have positive impact on physiological and behavioral...
https://www.frontiersin.org/research-topics/37517/interpretable-algorithmic-approaches-in-biomedical-signal-and-image-processing-for-enhanced-clinical About this Research Topic Biomedical data analysis is the most essential building block of biomedical engineering. The evolving techniques in signal and image processing have provided solutions to major healthcare problems across the world and improved diagnostic decision-making. Lately, due to the massive growth in biomedical data, there has been a considerable shift towards Machine Learning (ML) and Deep Learning (DL) based techniques. Despite their increased performance and value, these types of approaches tend to act as black-boxes and often lack transparency and explainability, raising concerns in both research and clinical practice. Explainability has become a major attribute nowadays since it can facilitate the understanding of an algorithm/model and lead to insights that can be utilized to improve the algorithm itself but also enhance clinical decision outcomes. With this collection, we want to give the opportunity to new and leading researchers to showcase their creative high-quality work on various explainable algorithmic approaches for biomedical data processing and analysis. Our main goal is to encourage the development of innovative techniques that can provide explainable results, as well as informative biomarkers for both diagnostic and research purposes. The complexity of biosignals/images requires first of all algorithms that take into account possible inherent nonstationarities/nonlinearities, increased noise levels, and/or inter- and intra-subject variability. Ideally, they should be autonomous (i.e., able to learn with minimal supervision), but most importantly, they should provide means to convey the relative importance of the features that they extract (i.e., human comprehensibility). One such example is model-driven learning, which is an emerging topic in signal/image processing. Model-driven learning fuses the disciplines of ML/DL with pure mathematical modeling in order to maintain both efficiency and interpretability (e.g. Laguerre Volterra Networks, Variable Projection Networks, Spiking Networks whereby network architecture (neurons, layers, connections, etc.) are inspired by mathematical models and the free parameters are considered to be trainable). However, other non ML/DL-based algorithms such as system identification or classical time-series techniques provide interpretable features (e.g, time and frequency domain features linked to specific physiological mechanisms). Novel perspectives of such approaches are welcome. This collection will include all manuscript types (i.e., Original Research, Review, Perspective articles, etc.). The topics of interest include (but are not limited to): • Brain-computer interfaces • Neuroengineering • Computational and network neuroscience • Neuroimaging • Wearable data processing Interpretable algorithmic approaches revolving around: • Time-series analysis • Time-frequency analysis • System identification • Connectivity analysis • Explainable machine learning/deep learning • Model-driven learning These are the main topics that we address in this Research Topic. Please note that the presented algorithms should be able to provide explainable results/features for either clinical practice or research. Analysis of various types of biosignals (e.g., EEG, MEG, ECG, EOG, PPG, extracellular and intracortical recordings, and other physiological signals) and images (e.g., ultrasound, (f)MRI, fNIRS) are highly welcome. Keywords: Interpretable algorithms, Biomedical Signal Processing, Image Processing, Machine Learning, Time-series analysis
One of the fundamental questions that I would like to address is whether we can predict long-term neurodevelopmental outcome from neuroimaging connectomic data acquired at an early time point. Potential early identification of newborns at high risk of developmental impairments could be tremendously valuable since it would allow a personalized and targeted early intervention window focusing on preventing/improving unfavorable outcomes. Furthermore, the characterization of biomarkers could provide insight into the underlying neural mechanisms with potential clinical usefulness.