Curriculum vitae - Projects & Research Interests - Publications - Fundings - Software tools - Students - Misc
In 2010, I initiated the CYTOMINE research project which lead to the development of the CYTOMINE open-source web software platform (Marée et al., Bioinformatics 2016; Rubens et al., 2019). Cytomine is a "Google/OpenStreet Maps"-like rich internet application for remote visualization, collaborative annotation and automated analysis of high-resolution, multi-gigapixels images.
It is now actively used in various domains (including digital pathology, large-scale microscopy, and other fields beyond the biomedical field) by various entities around the world collaborating over the web.
We are continuously developing new software modules and algorithms for multimodal data sources (I'm co-lead of the Software workpackage of the COMULIS COST ACTION), benchmarking, and applications in various fields (see CYTOMINE research project page). In addition to our ongoing research & development at University of Liège, I'm co-founder of Cytomine SCRL FS, a not-for-profit cooperative company established to ease usage and improve sustainability of this open-source tool.
I'm very much in favor of basic principles of open science (open access, open data, open source, open hardware) and links to commons (shared resources), e.g. I published this review on Open Practices and Resources for Collaborative Digital Pathology (Frontiers in Medecine, 2019), and we designed BIAFLOWS, a Collaborative Framework to Reproducibly Deploy and Benchmark Bioimage Analysis Workflows (Cell Patterns, 2020; built on top of Cytomine).
I'm also involved in the BigPicture project (2021-2027, deputy lead of Work Package 4 on "software tools") to establish the biggest open-access database of pathology images to accelerate the development of artificial intelligence in medicine.
Since 2003 we develop general-purpose methods for the recognition of various types of images that share some visual regularities, without relying on too strong assumptions about patterns to recognize and acquisition conditions, and without having to rely on domain experts to design specific features. This book chapter (2013) summarizes our previous work where we combined ensemble of randomized decision trees with random extraction of subwindows (square patches) described by their raw pixel values. To assess versatility of our methods, we often perform large-scale empirical studies e.g. in Pattern Recognition Letters (2016) for image classification, or Nature Scientific Reports (2018) in for interest point (landmark) detection. More recently, we are trying to combine ideas from tree-based methods, deep learning methods, and transfer learning approaches, see e.g. our papers in CVPR-CVMI (2018) and IEEE-JBHI (2020). Since 2016 these algorithms are integrated into the aforementioned CYTOMINE open-source software.
Between January 2005 and October 2014, I was the GIGA Bioinformatics platform manager. We offered software development and data analysis services to academic (within the GIGA research center and beyond) as well as to industrial researchers. Services included classification of biological/biomedical data (SELDI mass spectra, microarrays, clinical databases, ...) obtained from various medical instrumentation based on machine learning methods. This activity lead to co-authorship of several journal papers (in Journal of Immunology, Proteomics, Annals of the rheumatic diseases, ...) in various application domains (inflammatory diseases, cancer, ...).