Single-cell physical phenotyping of mechanically dissociated tissue biopsies for fast diagnostic assessment

Rapid and accurate histopathological diagnosis during surgery is critical for clinical decision-making. The prevalent method of intraoperative consultation pathology is time, labour and cost intensive and requires the expertise of trained pathologists. Here, we present an alternative technique for the rapid, label-free analysis of biopsy samples by sequentially assessing the physical phenotype of singularized, suspended cells in high-throughput. This new diagnostic pipeline combines enzyme-free, mechanical dissociation of tissues with real-time deformability cytometry at measurement rates of 100 - 1,000 cells/sec, and machine learning-based analysis. We show that physical phenotype parameters extracted from brightfield images of single cells can be used to distinguish subpopulations of cells in various tissues, without prior knowledge or the need for molecular markers. Further, we demonstrate the potential of our method for inflammatory bowel disease diagnostics. Using unsupervised dimensionality reduction and logistic regression, we accurately differentiate between healthy and tumorous tissue in both mouse and human biopsy samples. The method delivers results within 30 minutes, laying the groundwork for a fast and marker-free diagnostic pipeline to detect pathological changes in solid biopsies.