EPFL

Bidirectional generative framework linking single-cell nuclear morphology and gene expression, built on a morphology foundation model trained on 21M+ segmented nuclei.

A transformer that reasons over entire proteomes to produce context-aware protein representations for zero-shot protein-protein interaction and gene essentiality prediction.

A CLIP-based visual-language foundation model for multi-modal ophthalmic imaging, enabling zero-shot disease detection across 11 modalities including fundus, OCT, and slit-lamp.

Retrieval-augmented encoder-decoder that conditions pretrained ESM-2 on homologous sequences via cross-attention for improved masked prediction and conditional protein sequence generation.

A multimodal generalist foundation model for ophthalmic imaging, self-supervised on 2.78M images across 11 modalities for diagnosis, prognosis, and visual question answering.

Self-supervised 3D cell segmentation for fluorescence microscopy using WNet3D and Swin-UNetR, achieving supervised-level performance without annotated training data.