The application of Artificial Intelligence (AI) in radiology improves diagnostic accuracy and efficiency by automating routine tasks, detecting subtle patterns, and analyzing large datasets from various types of medical images, including X-rays, CT scans, MRIs, and mammograms.

The application of Artificial Intelligence (AI) in ophthalmology enhances diagnostic accuracy, reduces false positives, and enables personalized treatment plans by analyzing retinal images and identifying subtle changes indicative of various eye conditions.

Reproducible research not only leads to proper scientific conduct but also provides other researchers the access to build upon previous work. Most importantly, the person setting up a reproducible research project will quickly realize the immediate personal benefits: an organized and structured way of working. The person that most often has to reproduce your own analysis is your future self!

Mednet is disease classification and grading library for medical imaging based on PyTorch.

Sleepless is library for sleep phase detection from polysomnographs, based on Scikit-Learn and PyTorch.

This thesis develops stateless methods for sleep-phase detection from polysomnographs (PSG), while exploring techniques to improve cross-database generalisation.

In this thesis, we developed a technique to learn vessel segmentation from retinal fundus images using weakly-supervised methods.

Deepdraw is semantic segmentation library for medical imaging based on PyTorch.

In this thesis, we sought at developing precise neuron segmentation in order to support future study on the morphological phenotype of amyotrophic lateral sclerosis.

The course provides actionable guidelines on how to convert non-reproducible scripts with hard-coded parameters into reusable software.

Performance measures are an important component of machine learning algorithms. They are useful when it comes to evaluate the quality of a model, but also to help the algorithm improve itself. When used in small data sets, these measures may not properly express the performance of the model. That is when confidence intervals and credible regions can be useful. Expressing the performance measures in a probabilistic setting allows one to develop them as distributions. One can then use those distributions to establish credible regions.

An invited talk on the subject of fairness in Machine Learning applied to Computer-Aided Diagnosis at the University of Lausanne.

Short interview for a periodical publication from the COOP group in Switzerland, called Cooperation.

I contributed to a round table, part of the Industry Connect programme, sponsored by alp+ict, CimArk and Idiap (in French)

In this study, we investigate the benefits of automatic Pulmonary Tuberculosis (PTB) detection methods based on radiological signs found on frontal chest X-Ray images.

Article about my group's current work on the Swiss Digital Health Platform (in French)

My invited talk at University of Zurich on Reproducible Research background, motivations and methodology.

Bob is a signal-processing and machine learning toolbox originally developed by the Biometrics Security and Privacy Group, the Medical AI Group, and the Research and Development Engineers Group at the Idiap Research Institute, in Switzerland. Bob is primarily developed through GitLab.

This is an introductory course on Ethics and Reproducibility in Artificial Intelligence (AI). The course is composed of two parts. The first part covers ethical aspects of AI, while the second, practical aspects on building AI systems so they are continuously reproducible and extensible. It is given to master students at the Master in AI by the Idiap Research Institute, Switzerland.

This course, divided in two trimesters (modules M06 and M08), presents fundamental tools used in machine learning ranging from the most basic to more advanced. It is given to master students at the Master in AI by the Idiap Research Institute, Switzerland.

This course (EE-612) presents fundamental tools used in Machine Learning ranging from the most basic to more advanced. It is given to post-grad (Ph.D.) students at the École Polytechnique Fédérale de Lausanne, Switzerland.

The goal of this study is to identify whether it is possible to predict the occurrence of adverse events in patients based on their clinical data.

Early and accurate prediction of decompensation (functional deterioration) in patients in domestic settings may help prevent deaths. Based on values that can be measured from portable devices such as heart rate, blood oxygen saturation, systolic blood pressure, temperature, and age, we study the prediction capability of machine learning algorithms to determine patient decompensation (death) in the next 24 hours.

We address the problem of reproducible research in remote photo-plethysmography (rPPG). Most of the work published in this domain is assessed on privately-owned databases, making it difficult to evaluate proposed algorithms in a standard and principled manner. As a consequence, we present a new, publicly available database containing a relatively large number of subjects recorded under two different lighting conditions. Also, three state-of-the-art rPPG algorithms from the literature were selected, implemented and released as open source free software. After a thorough, unbiased experimental evaluation in various settings, it is shown that none of the selected algorithms is precise enough to be used in a real-world scenario.

I have actively worked in computer vision and deep learning (mostly) associated to biometric recognition, with potential application to various other tasks. Contributions range from the collection of datasets, the exploration of different methods to address and assess biometric recognition vulnerabilities, domain adaptation, and remote photoplethysmography.

My talk at the Valais/Wallis AI workshop on reproducible research

I was interviewed for a document on biometric vein recognition in the Swiss's RTS 12h45 news

The BEAT platform is a European computing e-infrastructure for Open Science proposing a solution for open access, scientific information sharing and re-use including data and source code while protecting privacy and confidentiality.

Introductory video to the BEAT Platform for Open Science and Reproducible Research

This is a course on Reproducible Research (RR) for research engineers working with software applications in Pattern Recognition (PR) and Machine Learning (ML). It motivates and explains concepts behind RR, an increasing trend in scientific publications in this niche, its implications and tools for implementing it on an individual or group levels. It is a hands-on course in the sense students will be required to create their own workflows for selected problems in ML and PR. By the end of this course, students should understand the basic concepts of reproducibility, its importance on their daily practice and how to achieve it with freely available tools and environments.

I was interviewed for a document in Jornal Nacional, one of the most viewed brazilian 20h00 news program.