José Dolz
Associate Professor
Software and Information Technology Engineering Department (LOGTI)
Office: A-3417
Phone: 514 396-8907
Jose.Dolz@etsmtl.ca
Currently working as Associate Professor at the ETS Montreal. My current research interests focus on medical image analysis and deep learning techniques with special interest on learning with limited supervision (e.g., weak supervision, few-shot learning).
Before that, I was post-doctoral researcher at the LIVIA department of the ETS in Montreal, under the supervision of Prof. Ismail Ben Ayed and Prof. Christian Desrosiers. My research as post-doc focused on bringing the power and strengths of deep learning to the problem of automatizing medical image segmentation. In addition to deep learning I explored some regularization techniques and optimization techniques to improve segmentation performance.
Previously, I worked as research engineer at AQUILAB at the same time that I was enrolled as PhD student at the scientific unit INSERM Onco-Thai U1189 “Interventional Therapies Assisted by Image and Simulation”.
Currently working on the intersection between computer vision, machine learning and medical image analysis. Research interests include image segmentation, learning with limited supervision (weakly supervised, semi-supervised, few-shot and unsupervised learning) and calibration of deep models.
Research interests:
- Health technologies
- Information and Communications Technologies
Research areas:
- Image segmentation
- Medical image analysis
- Machine learning
- Deep learning
- Weakly Supervised Learning
- Computer vision
‘+ Original articles in refereed journals and books chapters
«Determining regional brain growth in premature and mature infants in relation to age at MRI using deep neural networks»Beizaee, Farzad, Bona, Michele, Desrosiers, Christian, Dolz, Jose and Lodygensky, Gregory.” |
«An automatic deep learning-based system for screening and management of DME»Galdran, A., Chakor, H., Kabir, W., Kobbi, R., Liu, B., Dolz, J. and Ben Ayed, I..” |
«Calibrating segmentation networks with margin-based label smoothing»Murugesan, Balamurali, Liu, Bingyuan, Galdran, Adrian, Ayed, Ismail Ben and Dolz, Jose.” |
«Segmentation with mixed supervision: Confidence maximization helps knowledge distillation»Liu, Bingyuan, Desrosiers, Christian, Ben Ayed, Ismail and Dolz, Jose.” |
«Source-free domain adaptation for image segmentation»Bateson, Mathilde, Kervadec, Hoel, Dolz, Jose, Lombaert, Hervé and Ben Ayed, Ismail.” |
Read More >
+ Papers in refereed conference proceedings
«Mixup-privacy: A simple yet\xa0effective approach for\xa0privacy-preserving segmentation»Kim, Bach Ngoc, Dolz, Jose, Jodoin, Pierre-Marc and Desrosiers, Christian.” |
«Harmonizing flows: Unsupervised MR harmonization based on\xa0normalizing flows»Beizaee, Farzad, Desrosiers, Christian, Lodygensky, Gregory A. and Dolz, Jose.” |
«Constrained deep networks: Lagrangian optimization via Log-barrier extensions»Kervadec, H., Dolz, J., Yuan, J., Desrosiers, C., Granger, E. and Ben Ayed, I..” |
«The devil is in the margin: Margin-based label smoothing for network calibration»Liu, Bingyuan, Ben Ayed, Ismail, Galdran, Adrian and Dolz, Jose.” |
«Leveraging labeling representations in\xa0uncertainty-based semi-supervised segmentation»Adiga Vasudeva, Sukesh, Dolz, Jose and Lombaert, Herve.” |
‘
GitHub:
Jose
Dolz
Research & Innovation
Contact Us
Pavillon Principal (A)
1100, rue Notre-Dame Ouest
Montréal, QC, H3C 1K3
Room A-3600
Tel.: +1 (514) 396-8650
E-Mail: eric.granger@etsmtl.ca