Partial patients can’t benefit from current treatment strategy because of the existed tumor heterogeneity. While integrated analysis of imaging features, genetic molecules, and clinical phenotypes could be characterized tumor heterogeneity for making a more personalized treatment plan. However, there is a lack of exploration of cross-model database. To address these unmet needs, we established a database that includes 9,965 genes, 5,449 proteins, 1,121 metabolisms, 283 pathways, 854 imaging features and 73 clinical factors of colorectal cancer patients. This database demonstrates the correlation between genetic molecules, imaging features, with clinical factors, and predicting key genes expression and function by imaging feature. The MIPD database provides multi-omics and multi-modal data for improving the interpretability of cross-modal.