Imaging genomics, combining multi-modal imaging technologies with omics data, is evolving with the aid of AI and deep learning to decode complex disease mechanisms at molecular levels. This integrative approach provides insights for phenotypic and genomic correlations, notably enhancing cancer and cardiovascular disease research. Challenges remain in data harmonization, model interpretability, and computational demands, but foundational models are advancing cross-modal integration capabilities, heralding a new era of precise diagnostic and therapeutic strategies.