Philadelphia—A new artificial intelligence-powered tool called MISO (Multi-modal Spatial Omics) can examine data from extremely small pieces of tissue (some as small as 400 square micrometers, the width of five humans). can detect cellular characteristics of cancer. hair. The tool was built by researchers at the Perelman School of Medicine at the University of Pennsylvania. Analyze sets of data and apply insights to even the smallest spots on medical images. A new paper detailing MISO, published today in Nature Methods, says it could potentially allow doctors to derive the best personalized treatments for different cancers.
Using MISO, researchers used data and images from patient tissue provided to discover new information about a variety of cancers, including:
Bladder cancer: MISO has detected a specialized group of cells responsible for forming tertiary lymphoid structures associated with improved response to immunotherapy.
Stomach cancer: Distinguish between cancer cells and the mucous membrane inside the tissue using MISO
Colorectal Cancer: MISO has identified different subclasses of cancer cells that help shed light on the different malignant cells that make up a single tumor.
Additionally, MISO was used to analyze the structure of non-cancerous brain tissue.
All of these achievements can lead to better treatments and increase survival rates, but they will be extremely difficult, if not impossible, without very powerful AI tools like MISO.
MISO allows researchers to try to gain insight into different conditions by considering the physical arrangement of tissues by considering different “omics” modalities such as transcriptomics (the study of genes). It was developed to address the research field of “spatial multi-omics.” expression), proteomics (proteins), metabolomics (metabolites and their processes), etc.
“As the field of spatial omics advances, it will be possible to measure multiple omics modalities from the same tissue slice, providing complementary information and providing a more comprehensive and insightful view.” said. Dr. Mingyao Lisenior author of the study and professor of biostatistics and digital pathology. “MISO addresses the massive data challenge by enabling simultaneous analysis of all spatial omics modalities and, when available, microscopic anatomical images. It is the only method that can handle such datasets that include cells.”
When viewing images using spatial transcriptomics, a single pixel within an image contains 20,000 to 30,000 data points that are analyzed through an omics lens, and multiple omics are taken into account. If so, that number could double or triple. In MRI and CT scans, there is only one data point (shade of gray) per pixel to interpret. Without some kind of artificial intelligence tool to assist them, it would be nearly impossible for doctors and researchers examining medical images to realize some of the insights that MISO can provide.
The latest efforts in omics image technology
MISO continues Lee’s work in developing artificial intelligence imaging technology that can see things that even trained humans cannot see. Earlier this year, her team Paper investigating iSTARa tool to explore genomics, has discovered traces of cancer and the body’s response to its treatment that would otherwise go unnoticed.
Although MISO considers a much wider range of data than iSTAR, and some of its components were even used in the development of MISO, Li believes both are useful but in different areas. iSTAR will help improve imaging clarity and virtually generate spatial-omics data that MISO can analyze. MISO, on the other hand, supports insight into more subtle targets (such as the detection of highly endothelial venules, special cell groups that recruit white blood cells to specific tissues).
Going forward, the team hopes to gather everything they’ve learned about spatial omics and pathology imaging and improve MISO so that it can simultaneously analyze multiple tissue samples at once, dramatically increasing the output of results.
Some data, such as epigenetic marks (chemicals that control DNA but are influenced by the environment rather than pure genetics), have not yet been measured at scale, but MISO’s AI system will This is because it can “learn” as it processes the information, so that it can recognize this data as it becomes available. “We expect that by integrating these diverse data types, MISO will be able to provide deeper insights into different aspects of cell behavior,” Lee said.
This research was supported by National Institutes of Health grants (R01HG013185 and R01LM014592).
Penn Medicine is one of the world’s leading academic medical centers, dedicated to the related missions of medical education, biomedical research, excellence in patient care, and community service. The organization consists of: University of Pennsylvania Health System and Mr. Penn’s Raymond and Ruth Perelman School of MedicineIt was established in 1765 as the first medical school in Japan.
Perelman School of Medicine consistently ranks among the nation’s top recipients of National Institutes of Health funding, with $550 million awarded in fiscal year 2022. With a proud history of “firsts” in medicine, the Penn Medicine team continues to advance modern technology, including recent breakthroughs such as CAR T-cell therapy for cancer and mRNA technology used in COVID-19 vaccines. It has pioneered discoveries and innovations that have shaped medicine.
University of Pennsylvania Health System’s patient care facilities stretch from the Susquehanna River in Pennsylvania to the coast of New Jersey. These include Hospital of the University of Pennsylvania, Penn Presbyterian Medical Center, Chester County Hospital, Lancaster General Hospital, Penn Medicine Princeton Health, and Pennsylvania Hospital, the nation’s first hospital, founded in 1751. Other facilities and companies include Good Shepherd Penn Partners; , Penn Medicine at Home, Lancaster Behavioral Health Hospital, Princeton House Behavioral Health, and more.
Penn Medicine is an $11.1 billion company supported by more than 49,000 talented faculty and staff.
