Researchers at the Children's Hospital of Philadelphia have created an artificial intelligence tool called CelloType.
This tool aims to improve the identification and classification of cells within complex tissue samples, an advancement that could transform the understanding of diseases like cancer and chronic kidney disease. The study detailing CelloType was published in the journal Nature Methods.
According to a Nov. 25 news release, CelloType is designed to address the growing need for advanced tools in the field of spatial omics. Spatial omics combines molecular profiling with spatial data to map the location and interaction of molecules within cells. This process provides a deeper understanding of how diseases develop at the cellular level, potentially leading to more precise diagnostics and treatments.
CelloType uses a type of AI called transformer-based deep learning. Unlike traditional two-step processes for cell segmentation and classification, CelloType integrates these tasks into a unified framework, making the analysis more efficient and accurate, according to the release. The tool also outperformed existing methods in detecting and categorizing cells across a variety of imaging techniques, including fluorescence and bright light images.
By analyzing multiplexed tissue images — advanced biomedical images that highlight multiple biomarkers — CHOP researchers were able to demonstrate that CelloType excels in segmenting and classifying both small and large structures within tissues.
CelloType has been made available as an open-source tool for noncommercial use.