A NEW open-source computational pipeline called MARQO enables single-cell analysis of whole-slide cancer tissue images in minutes, offering unprecedented detail and efficiency for investigating cellular organisation and marker expression in tumours. This technology promises to advance biomarker research and facilitate more personalised approaches in cancer care.
Cancer pathologists have long relied on immunostaining and microscopy to identify immune cells and molecular markers within tumour tissue. Traditional image analysis is fragmented and labour-intensive, often restricted to smaller tissue samples and requiring complex patchwork processing or costly computing clusters. The MARQO platform, developed at the Icahn School of Medicine at Mount Sinai, aims to transform this workflow by automating start-to-finish multiplex imaging analysis with a simple graphical interface.
Researchers tested MARQO on human tumour and adjacent normal tissue samples, using multiplex immunohistochemical and immunofluorescence stains. The pipeline processed full-slide images and successfully performed elastic registration, nuclear segmentation and unsupervised clustering with mini-batch k-means, enabling user-guided cell classification. MARQO’s results were validated against manual pathologist readings and quantified for multiple markers, revealing strong agreement and reproducibility. Its versatility was demonstrated across slides of diverse sizes, including biopsies and tissue microarrays, and with various staining protocols ranging from standard immunohistochemistry to twenty-colour multiplex panels. Notably, MARQO identified clinically relevant patterns, such as CD8+ T cell enrichment in hepatocellular carcinoma patients responding to immunotherapy in a phase 2 trial, showing its ability to pinpoint cellular and spatial mechanisms in situ at single-cell resolution.
For clinical practice, while MARQO is not yet accredited for diagnostics, its compatibility with clinical staining approaches and rapid output could soon make it a valuable tool for pathologists and researchers. This platform is poised to accelerate biomarker discovery and treatment prediction, supporting the shift toward precision oncology. Ongoing enhancements will add spatial neighbourhood analysis and high-performance computing integration, broadening its impact in cancer research and diagnostics.
Reference
Buckup M et al. Multiparametric cellular and spatial organization in cancer tissue lesions with a streamlined pipeline. Nature Biomedical Engineering. 2025;DOI:10.1038/s41551-025-01475-9.
