Multi-modal AI for personalised treatment of oesophageal cancer

-Deep Learning in digital pathology imaging with Python and PyTorch-Representation Learning with Deep Generative Models-Self-Supervised LearningPublication of a paper: Creating small but meaningful representations of digital pathology images, Corentin Guerendel, Phil Arnold, Ben Torben-Nielsen, Proceedings of the MICCAI Workshop on Computational PathologyPatent:Machine learning based medical data checker, Bejamin Torben-Nielsen, Fernando Garcia-Alcalde, Vincent Mauduit, Corentin Guerendel, Phil Arnold

Predicted cardiovascular diseases from fundus imaging: -Retinal vessel segmentation and artery/vein (A/V) classification using Deep Learning techniques with Python and Keras/Tensorflow: U-Net network, data augmentation, model averaging, test-time augmentation, etc.-Preprocessing: retinal images enhancement based on luminosity and contrast adjustment -Determination of arteriovenous ratio

-Digital phenotyping project to develop new plants -Image analysis in R and Python : object detection, image reconstruction, segmentation, etc.-Data preprocessing

-Developed a new statistical model to evaluate the quality of record linkage -Implemented this method in SAS -Wrote a user guide related to SAS programs dealing with record linkage

-Designed and programmed questionnaires in VBA -Implemented an error management program in VBA -Launched an email campaign

-Managed a team of 20 people -Managed a 25k€ budget -Organized sailing events -Found sponsors : +320% in comparison with the previous year