Enzo Ferrante
Date: May 15, 2026, 11 am
Location: Medical University of Vienna, Anna Spiegel Research Building, Seminar Room Level 3
Speaker: Enzo Ferrante
Title: Towards robust anatomical segmentation of medical images
Abstract: Deep learning models for medical image segmentation typically optimize pixel-level objectives, making it difficult to enforce global constraints on shape, topology, and spatial coherence. In this talk, I will present our efforts towards building robust and anatomically plausible segmentation models along two interconnected research axes.
First, I will introduce hybrid graph neural network architectures — HybridGNet (Gaggion et al., MICCAI, 2021; IEEE TMI, 2022) and its 3D extension HybridVNet (Gaggion et al., Medical Image Analysis, 2025) — that combine convolutional encoders with graph-based decoders to produce landmark-based segmentations with built-in topological guarantees. I will discuss image-to-graph skip connections for improved robustness under occlusion, differentiable rasterization for training with pixel-level annotations, the emergence of implicit anatomical correspondences, and uncertainty quantification via variational formulations (Cosarinsky et al., MIDL, 2026). I will also show how mesh-based cardiac representations enable imaging genetics, where unsupervised geometric deep learning on left-ventricular meshes from the UK Biobank led to the discovery of novel genetic loci associated with cardiac morphology (Bonazzola et al., Nature Machine Intelligence, 2024).
Second, I will briefly discuss fairness and bias in medical image analysis (Larrazabal et al., PNAS, 2020). Building on CheXmask, our dataset of over 676,000 chest X-ray segmentation masks (Gaggion et al., Scientific Data, 2024), I will present methods for unsupervised bias discovery based on Reverse Classification Accuracy (Gaggion et al., MICCAI FAIMI Workshop, 2023) that anticipate performance disparities across demographic subgroups without ground-truth annotations, including a recent extension using in-context segmentation models and conformal prediction (Cosarinsky et al., IEEE TMI 2026).
- Bonazzola, R., Ferrante, E., Ravikumar, N., et al. (2024). Unsupervised ensemble-based phenotyping enhances discoverability of genes related to left-ventricular morphology. Nature Machine Intelligence, 6, 291–306.
DOI: https://doi.org/10.1038/s42256-024-00801-1 - Larrazabal, A. J., Nieto, N., Peterson, V., Milone, D. H., & Ferrante, E. (2020). Gender imbalance in medical imaging datasets produces biased classifiers for computer-aided diagnosis. Proceedings of the National Academy of Sciences, 117(23), 12592–12594.
DOI: https://doi.org/10.1073/pnas.1919012117 - Cosarinsky, M., Billot, R., Mansilla, L., Jimenez, G., Gaggion, N., Fu, G., Tirer, T., & Ferrante, E. (2026). ConfIC-RCA: Statistically grounded efficient estimation of segmentation quality. IEEE Transactions on Medical Imaging. Advance online publication / accepted in press.
DOI: https://doi.org/10.1109/TMI.2026.3679618
Bio: Dr. Ferrante completed his PhD in Computer Science at Université Paris-Saclay and INRIA (Paris, France), carried out postdoctoral research at Imperial College London (UK), and earned a Systems Engineering degree from UNICEN University (Tandil, Argentina). He has also been a visiting PhD student at Stanford University, a Fulbright Visiting Researcher at Harvard Medical School in Boston, an Invited Professor at Université Paris-Saclay, France, and currently a Von Humboldt Visiting Researcher at the Technical University of Munich.
His research interests span machine learning for computer vision and NLP, with a current focus on fairness and robustness in healthcare applications and biomedical imaging.
He is a Staff Research Scientist leading a group at the Applied Artificial Intelligence Lab of Argentina's National Research Council (CONICET) and University of Buenos Aires; the Academic Leader of AnyoneAI, an EdTech startup focusing on creating AI talent in LATAM; and the Head of Machine Learning at Apolo Biotech, a startup focusing on RNA design to boost crop production. His research has been recognized with several awards, such as the Google Award for Inclusion Research, the Distinguished International Associate Award from the UK Royal Academy of Engineering, and the Friederich Wilhelm Bessel Award from the Von Humboldt Foundation, among others.
Aarti Krishnan
Date: May 15, 2026, 11:45 am
Speaker: Aarti Krishnan
Tittle: A generative deep learning approach to de novo antibiotic design
Abstract: Antimicrobial resistance is creating an urgent need for antibiotics with fundamentally new structures. Although deep learning methods can identify antibacterial compounds from existing chemical libraries, they often yield molecules with limited structural novelty. To address this challenge, we developed a generative AI framework for de novo antibiotic design using two complementary strategies, both built on genetic algorithms and variational autoencoders. First, we used a fragment-based approach, screening more than 45 million chemical fragments in silico against two priority pathogens, Neisseria gonorrhoeae and Staphylococcus aureus, and then expanding the most promising hits. Second, we used an unconstrained de novo design approach.
Of the 24 compounds we synthesized, seven showed selective antibacterial activity. Two lead compounds were especially promising: they demonstrated bactericidal efficacy against multidrug-resistant clinical isolates, acted through distinct mechanisms, and reduced bacterial burden in mouse models of N. gonorrhoeae and methicillin-resistant S. aureusinfection. We also validated structurally related analogs in both compound classes, supporting the broader potential of these scaffolds as antibacterial agents. Together, these findings show that generative deep learning can guide the design of novel antibiotics and provide a platform for exploring previously uncharted regions of chemical space. https://www.cell.com/cell/abstract/S0092-8674(25)00855-4
Bio: Aarti Krishnan is a postdoctoral associate in the Collins Lab at MIT and the Broad Institute of MIT and Harvard, where her research bridges computational and experimental approaches to drug discovery, with a particular focus on antimicrobial resistance. She has applied generative AI algorithms and structural tools such as AlphaFold to predict antibacterial mechanisms of action and advance the development of new, safe antibiotics.
Originally from southern India, Aarti moved to Switzerland for her Master's in Computational Biology from ETH Zürich and her Ph.D. in Biomedical Sciences from the University of Geneva, where she developed and validated a metabolic network model of an intracellular eukaryotic parasite to uncover and target novel therapeutic vulnerabilities. Her interdisciplinary work spans deep learning, systems biology, infectious disease, and microbial metabolism.
This is part of the CAIM Talks series.