The rapid growth of digital pathology and advances in self-supervised deep learning have enabled the development of foundational models for various diseases. While multimodal approaches integrating diverse data sources have emerged, a critical gap remains in comprehensively integrating whole-slide imaging (WSI) and spatial transcriptomics (ST). This is essential for capturing significant molecular heterogeneity beyond standard hematoxylin & eosin (H&E) staining. In this paper, we introduce SPADE, a foundational model that integrates morphological and molecular information to drive image representation learning in a single latent space. SPADE leverages a mixed-data expert technique, where experts are generated through two-step image feature spatial clustering, to learn representations of co-registered WSI patches and gene expression profiles using contrastive learning. Pretrained on the comprehensive HEST-1k dataset, SPADE is evaluated on 20 downstream tasks, demonstrating significantly better small-shot performance than baseline models, highlighting the benefits of integrating morphological and molecular information in a single latent space.
