A new study led by the Bodenmiller Lab presents CISI-IMC, a compressed sensing framework that doubles the effective multiplexity of imaging mass cytometry. By leveraging protein co-expression patterns and computational decompression, the method reconstructs high-dimensional protein data from fewer imaging channels. Using a training dataset spanning multiple tissue types, the team developed a universal expression dictionary and barcoding matrix.
Tested across 71 tumor and healthy tissues, CISI-IMC achieved a mean Pearson correlation of 0.8 when compared to ground-truth measurements, maintaining high accuracy in cell type classification. This advance lays the groundwork for expanding IMC beyond the current 40-channel limit, with potential scalability to 80 markers or more.
The work highlights DQBM’s commitment to enabling next-generation spatial biology through interdisciplinary innovation.
DOI: 10.1038/s41467-025-66629-4