CellsiLAB: Cells and Image Analysis
Biomedical image analysis area of the Codalab research group at the Universitat Politècnica de Catalunya (UPC).
We work on applied artificial intelligence, focused mainly (though not exclusively)
on biomedical image analysis. We develop deep learning and
computer vision methods to support clinical decision-making, with a strong emphasis on
explainable and trustworthy AI: we care not only about what
a model predicts, but why.
Research lines
- Explainable AI (xAI) for biomedical imaging: new methodologies, evaluation strategies and clinical applications.
- Computer vision for medical imaging: classification, segmentation and quality assessment.
- Generative and vision-language models applied to medical data.
- Federated learning for robust multi-center generalization.
Application domains
- Hematology: blood cell classification, myelodysplastic syndromes, leukemia and red cell morphology.
- Obstetrics: fetal ultrasound quality assessment (nuchal translucency).
- Pediatric ophthalmology: retinopathy of prematurity.
Research projects
- Explainable deep learning in medical image analysis: novel methods, evaluation strategies and clinical applications (PID2023-146261OB-I00, 2024 to 2027).
- Computational system for the diagnosis of acute leukemia and lymphoma from peripheral blood images, including a proof of concept and technological valorization (PDC2022-133514-I00, 2022 to 2024).
- CellsiMaticDeep, computational hematology: deep learning solutions for the diagnosis of hematological diseases from peripheral blood cell images (PID2019-104087RB-I00, 2020 to 2023).
- Characterization and automatic classification of leukemic cells by means of digital image processing and pattern recognition for diagnosis support (DPI2015-64493-R, 2016 to 2018).
Institutions and collaborations
- Universitat Politècnica de Catalunya (UPC)
- Hospital Clínic de Barcelona, CORE Laboratory
- Hospital Sant Joan de Déu, Barcelona
- Centro de Diagnóstico Biomédico (CDB)
- International partners across Europe, Latin America and Asia
Team
- Dr. José Rodellar. Professor (Emeritus), UPC. Signal processing and machine learning for blood cell morphology. Profile
- Santiago Alférez. Assistant Professor, Dept. of Mathematics, UPC. Machine learning, statistics and explainable AI for medical image analysis. Profile
- Kevin Barrera. Assistant Professor, UPC. Deep learning for cell morphology. Profile
- Dr. Anna Merino. Clinical lead, CORE Laboratory, Hospital Clínic de Barcelona. Hematology and cell morphology.
Datasets and Models
Our public datasets and models will be published here soon. Stay tuned.
Links
For collaborations or inquiries, please reach out through our
website.