Deep learning-enabled detection of rare circulating tumor cell clusters in whole blood using label-free, flow cytometry

Abstract

We present a deep-learning enabled, label-free flow cytometry platform for identifying circulating tumor cell clusters in whole blood based on the endogenous scattering detected at three wavelengths. The method has potential for in vivo translation.

The Role of Circulating Tumor Cells as a Liquid Biopsy for Cancer: Advances, Biology, Technical Challenges, and Clinical Relevance

“…This process, known as angiopellosis, allows them to retain their multicellular phenotype and effectively proliferate at distant sites, a phenomenon central to the Cancer Exodus Hypothesis [11,76,77,[81][82][83]. Advanced imaging and isolation techniques have been instrumental in observing these processes, revealing that CTC clusters exhibit unique molecular profiles that enhance their metastatic capabilities [13,31,[83][84][85][86].…”

“…AI algorithms can process and analyze vast amounts of data from CTCs, identifying patterns and biomarkers that might be missed by traditional methods. Machine-learning models can be trained to recognize specific CTC characteristics, such as morphological features, genetic mutations, and protein expressions, enhancing the accuracy and efficiency of CTC detection and characterization [85,106,109].…”

“…Microfluidic devices allow for the precise manipulation of fluids at a micro-scale, improving the isolation and capture of CTCs. Nanotechnology, particularly the use of nanoparticles, can enhance the sensitivity of CTC detection by targeting specific tumor markers or properties [13,25,26,31,34,85].…”