New tool combines multiple data sources to detect tissue structures in cancer
FEBRUARY 2025
A team of Emory researchers has developed a new artificial intelligence-powered tool for analyzing tissue samples called MISO (Multi-modal Spatial Omics) that can surpass the abilities of expert human pathologists. MISO detects immune cells and subtle variations in tumor biopsies. It allows researchers to integrate several types of data, such as histology images and measurements of gene expression, and metabolic activity within cells.
A paper describing MISO's performance across multiple tissue types was published in Nature Methods. The tool results from a collaboration between Jian Hu, PhD, assistant professor of human genetics at Emory, and researchers at the University of Pennsylvania. "MISO enables researchers to integrate multi-modality molecular measurements with morphological imaging," Dr. Hu explains. "It can handle diverse multi-omics datasets of varying quality and uncover novel biological insights that cannot be revealed when analyzing each modality individually."
The paper includes an analysis of MISO's segmentation of a colon cancer tissue sample, as well as MISO's ability to detect tertiary lymphoid structures within bladder cancer biopsies. The presence of tertiary lymphoid structures within a tumor is linked with response to cancer immunotherapy. Beyond cancer pathology, the authors say MISO could be used to analyze and classify brain cells in a tissue sample.
Citation: Coleman et al. (2025). Resolving tissue complexity by multimodal spatial omics modeling with MISO.