Talks

Presentation of the MARIO challenge results — an AMD progression monitoring benchmark in multi-modal OCT imaging, organised as a MICCAI 2024 satellite event.

We introduce LatiM, a continuous-time longitudinal representation learning model for predicting disease progression from irregular sequential medical images.

Self-supervised longitudinal learning via Neural ODEs applied to retinal image sequences for diabetic retinopathy follow-up and progression prediction.

LMT enables predicting disease progression from a single image at inference time by leveraging longitudinal data mixing during training across multiple sources.

Longitudinal self-supervised learning approach applied to retinal fundus image sequences for automated diabetic retinopathy detection. Awarded Best Paper.