Diana Mateus
Centrale Nantes
and
the SIMS team from the LS2N lab.
Since 2018 I have been the holder of the MILCOM Chair. MILCOM aims to design machine-learning methods that explicitly consider the challenges of analysing medical images such as dealing with volumetric multi-modal and heterogenous data, small and imbalanced databases, and/or limited access to expert annotations.
Google Scholar Profile
LinkedIn Profile
Early Breast Detection from Images
This project points to the early breast cancer detection from mammographies. The challenges are the heterogeneous and incomplete nature of the annotations. The main contribution has been the proposition of a self-supervised approach to guide a convolutional neural network to detect and spatially locate anormalities while training with image-wise annotations only\footnote{\bibentry{tardy2021abnormalities}}. We have also explored two approaches to measure the uncertainty of the predictions \footnote{\bibentry{tardy2019uncertainty}} and a multi-task method to learn from hetereogenous labels\footnote{\bibentry{tardy2022multitask}}. These are outcomes of Mickael Tardy’s Ph.D.
A second Ph.D., Guillaume Pelluet, follows this fruitful industrial collaboration with the company Herami. He has focused on the multi-scale nature of the problem\footnote{\bibentry{pelluet2021brainles}} and on further increasing the interpretability of the results through graph CNNs\footnote{\bibentry{pelluet2022miua}}.
-
Mickael Tardy and Diana Mateus. Looking for abnormalities in mammograms with self- and weakly-supervised reconstruction. IEEE Transactions on Medical Imaging (TMI), January 2021. (JCR)
-
Mickael Tardy, Bruno Scheffer, and Diana Mateus. Uncertainty measurements for the reliable classification of mammograms. In International Conference on Medical Image Computing and Computer Assisted Interventions (MICCAI), pages 495–503. Springer, Cham, October 2019
-
Mickael Tardy and Diana Mateus. Leveraging multi-task learning to cope with poor and missing labels of mammograms. Frontiers in Radiology, 1, 2022
-
Guillaume Pelluet, Mira Rizkallah, Oscar Acosta, and Diana Mateus. Unsupervised multimodal supervoxel merging towards brain tumor segmentation. In Brain Lesion workshop at MICCAI. Springer, 2021
-
Guillaume Pelluet, Mira Rizkallah, Mickael Tardy, Oscar Acosta, and Diana Mateus. Multi-scale graph neural networks for mammography classification and abnormality detection. In Medical Image Understanding and Analysis, pages 636–650, Cham, 2022. Springer International Publishing. (2nd Best paper award)