Researchers at the Children's Hospital of Philadelphia (CHOP) have developed an innovative artificial intelligence model called CelloType, which is designed to enhance the analysis of biomedical imaging. By integrating segmentation and classification tasks, this tool offers detailed insights into disease progression at the cellular level and could improve diagnostics and targeted treatments. Now publicly available for noncommercial use, CelloType represents a major step forward in the application of AI in precision medicine.
A Multitask AI Model for Biomedical Imaging
As Healthcare IT News reports, CelloType employs a multitask learning strategy that simultaneously integrates cell segmentation and classification tasks, setting it apart from traditional models that separate these processes. This novel approach improves accuracy and efficiency in detecting and classifying cells, especially those with irregular shapes or sizes—longstanding challenges in the field.
The researchers, whose work was funded by the National Institutes of Cancer and published in Nature Methods, demonstrated that CelloType outperforms other imaging models, including Mesmer and Cellpose2, in segmenting multiplexed tissue images. By leveraging transformer-based deep learning, CelloType can analyze high-dimensional data, uncovering complex relationships and context within tissue samples.
"Unlike the traditional two-stage approach of segmentation followed by classification, CelloType adopts a multitask learning strategy that integrates these tasks, simultaneously enhancing the performance of both," the authors wrote in their report. Lead author Kai Tan, MD, a professor in the Department of Pediatrics at CHOP, stated this approach “could redefine how we understand complex tissues at the cellular level, paving the way for transformative breakthroughs in healthcare."
Addressing the Needs of Spatial Omics
The development of CelloType addresses a growing need in spatial omics—a field that integrates genomics, transcriptomics, and proteomics with spatial data to map molecular locations within cells. Advances in this field are critical for understanding disease mechanisms, enabling precision diagnostics, and tailoring personalized treatments.
As an open-source tool, CelloType provides researchers and clinicians with access to cutting-edge technology that can improve the analysis of tissue samples in complex diseases like cancer and chronic kidney disease. Its ability to analyze high-dimensional data faster and more accurately than existing models positions it as a potential game-changer in diagnostics and therapeutic development.
CHOP’s innovation also reflects a broader trend of integrating AI into biomedical imaging. For example, in Norway and Denmark, mammography images are being used to predict breast cancer diagnoses, while Stamford Health has incorporated AI to screen for coronary artery disease during routine CT scans. These tools highlight how AI can improve healthcare outcomes by enhancing diagnostic capabilities and enabling earlier interventions.
Kai Tan summarized the significance of the technology: "We are just beginning to unlock the potential of this technology."
By making CelloType open source, CHOP has made an important contribution to advancing spatial omics and precision medicine. This tool has the potential to accelerate research, enhance disease understanding, and drive improvements in patient care.
