Summary: Researchers at the University of Pennsylvania have developed MISO, an AI tool that analyzes spatial -omics data from extremely small tissue samples, providing detailed cellular insights for cancer detection, treatment, and imaging advancements.
Key Takeaways
- Enhanced Cancer Detection and Analysis: MISO can analyze extremely small tissue samples and spatial -omics data, providing detailed cellular insights for cancers like bladder, gastric, and colorectal, as well as non-cancerous tissues.
- Integration of Multiple Modalities: The tool simultaneously processes data from transcriptomics, proteomics, and metabolomics, offering a comprehensive view of tissue structures and addressing challenges in handling datasets with hundreds of thousands of cells.
- Applications Beyond Current Imaging Techniques: By analyzing up to 30,000 data points per image pixel, MISO enables insights beyond the capabilities of traditional imaging methods like MRI or CT, with potential for large-scale studies and integration of emerging data types like epigenetic marks.
—————————————————————————————————————————————————————————
Researchers at the Perelman School of Medicine at the University of Pennsylvania have developed MISO (Multi-modal Spatial Omics), an artificial intelligence tool capable of analyzing extremely small tissue samples, some as small as 400 square micrometers. The findings, published in Nature Methods, highlight MISO’s ability to apply insights to medical imaging and improve understanding of cancer at the cellular level.
Understanding Cancer at the Cellular Level
MISO enables researchers to analyze spatial -omics data, which integrates transcriptomics, proteomics, metabolomics, and other modalities to provide detailed views of tissue structure.
“MISO addresses a huge data challenge by enabling simultaneous analysis of all spatial -omics modalities, as well as microscopic anatomy images when available,” says Mingyao Li, PhD, the study’s senior author and a professor of biostatistics and digital pathology. “It is the only method that is able to handle datasets like these with hundreds of thousands of cells per sample.”
The tool has already been applied to several cancer types:
- Bladder cancer: MISO identified cells forming tertiary lymphoid structures, associated with improved responses to immunotherapy.
- Gastric cancer: It distinguished between cancer cells and surrounding mucosa.
- Colorectal cancer: MISO identified different malignant sub-classes within a single tumor.
MISO was also used to study non-cancerous brain tissue structures, showcasing its range of applications.
Spatial -omics data provides a level of detail not possible with traditional imaging, such as MRI or CT, which rely on grayscale data. When analyzing spatial transcriptomics, a single image pixel may contain 20,000 to 30,000 data points, allowing MISO to detect patterns beyond human recognition.
Broader Applications
The development of MISO builds on earlier work with iSTAR, another AI tool focused on genomics. While iSTAR enhances imaging sharpness, MISO is designed to process complex datasets for detailed insights.
Future goals include improving MISO to analyze multiple tissue samples at once, increasing its capacity for large-scale studies. Additionally, MISO’s adaptable AI could integrate new data types, such as epigenetic marks, to further expand its utility. “I anticipate that integrating these diverse data types will enable MISO to provide deeper insights into various aspects of cellular behavior,” Li adds.
