Journal of Biophotonics, Journal Year: 2025, Volume and Issue: unknown

Published: April 7, 2025

ABSTRACT Colorectal cancer (CRC) is one of the most prevalent gastrointestinal malignancies, necessitating study cellular and molecular changes within tumor microenvironment. While pathological image analysis remains gold standard, its labor‐intensive nature limits broad application. This proposes a label‐free CRC typing approach using intelligent optical time‐stretch (OTS) imaging flow cytometry combined with multi‐instance learning. Specifically, we construct high‐throughput cell acquisition system by integrating OTS microfluidic focusing, capturing 363 931 images from 10 clinical samples. To address diversity heterogeneity, employ learning framework, which incorporates multi‐level attention mechanism to explore feature interactions at both channel instance levels. Finally, apply majority voting enable efficient typing. Our method achieves an accuracy 85.78% in distinguishing normal cancerous cells, while encouraging performance across all

Language: Английский