Related Experiment Video
Updated: Jun 24, 2026
Morphology-Based Distinction Between Healthy and Pathological Cells Utilizing Fourier Transforms and Self-Organizing Maps
Published on: October 28, 2018
Deep learning-enabled single-cell morpholomic atlas of nasal swabs distinguishes chronic inflammation from sinonasal
Brittany T Rupp1, Andreja Jovic2, Theresa Weaver1
1Department of Oral and Craniofacial Molecular Biology, Philips Institute for Oral Health Research, Virginia Commonwealth University, Richmond, VA, USA.
Abstract:
Sinonasal malignancies frequently present with symptoms overlapping chronic inflammatory conditions, complicating early detection and delaying treatment. A fast, non-invasive approach capable of resolving cell states across inflammatory and malignant disease from routine nasal swabs could substantially improve clinical screening. Using the REM-I platform, we generated a reference atlas of deep learning-enabled single-cell morpholomic features from >641K immune cell brightfield images. This reference atlas was integrated with >2.5 M images obtained from nasal swabs spanning health, chronic rhinosinusitis (CRS), and sinonasal carcinoma to perform differential feature testing and comparative feature enrichment across disease states. Sinonasal carcinoma samples exhibited distinct immune remodeling, including increased myeloid-enriched cell abundance and elevated small dark pixel intensity consistent with enhanced granulocyte activity. Basophil/NK-enriched clusters and epithelial clusters contained tumor-associated cells with deep learning-derived morphologic signatures not observed in CRS or healthy samples. These findings support the potential use of nasal swabs in early point-of-care diagnostics.
