Title:	Deep Learning for imaging and omics integration
DNr:	Berzelius-2025-380
Project Type:	LiU Berzelius
Principal Investigator:	Gisele Miranda <gmirand@kth.se>
Affiliation:	Kungliga Tekniska högskolan
Duration:	2025-11-11 – 2026-06-01
Classification:	10210
Keywords:

Abstract

Recent advances in cellular imaging and molecular profiling have transformed our ability to observe biological systems at unprecedented resolution and scale. Technologies such as high-content microscopy, spatially resolved transcriptomics, and single-cell omics now generate rich, complementary views of cellular organization and function. Yet, these diverse measurements remain fragmented across experiments, modalities, and scales, posing a fundamental challenge for building integrated models of cellular function. Our research group works at the intersection of omics and imaging data, developing machine learning frameworks that unify molecular profiles with spatial and morphological information. We aim to design interpretable and transferable representations that connect cellular morphology, dynamics, and molecular identity, enabling a holistic understanding of biological processes in health and disease. By combining predictive and generative approaches, from self-supervised representation learning to probabilistic modeling, we seek to build scalable, biologically informed models capable of simulating, predicting, and interpreting cellular behavior across diverse experimental contexts. Ultimately, our work bridges molecular and spatial biology through multimodal learning, establishing a foundation for data-driven discovery and translational insights.