Circulating tumor cell detection in cancer patients using in-flow deep learning holography

Area of Science:

• Biomedical Engineering
• Cancer Research
• Microfluidics

Background:

• Circulating tumor cells (CTCs) are vital biomarkers for cancer management but are difficult to detect due to low numbers and heterogeneity.
• Existing CTC detection methods face challenges in sensitivity and specificity, potentially missing crucial cell populations.

Purpose of the Study:

• To develop and validate a streamlined system for enhanced CTC enumeration using digital holographic microscopy (DHM) and deep learning.
• To improve the sensitivity and specificity of CTC detection, particularly for EpCAM-negative cells.

Main Methods:

• Integration of inertial microfluidics for cell enrichment with dual-modality imaging (holography and fluorescence).
• Application of a deep learning model for real-time, cell-by-cell morphological analysis.
• Supplementation with immunofluorescent profiling for improved enumeration accuracy.

Main Results:

• Demonstrated significantly higher CTC counts in late-stage prostate cancer patients versus healthy controls.
• Achieved a low patient-level false positive rate of 1 cell/mL.
• Identified a substantial proportion of CTCs as EpCAM-negative but PSMA-positive, challenging traditional detection markers.

Conclusions:

• DHM-based systems offer a powerful, label-free method for high-throughput CTC identification.
• The proposed platform enhances CTC detection sensitivity and specificity, especially for challenging cell types.
• Findings underscore the limitations of EpCAM as a sole marker and highlight DHM's potential in cancer screening, diagnostics, and precision oncology.