Abstract

Background

Extravasation, invasion, epithelial-to-mesenchymal transitions, metastasis progression, immune evasion, and therapeutic resistance are driven by phenotypic alterations in cancer cells. Assessing cell morphology, stiffness, and deformability is therefore crucial. The expression of PD-L1, HER2, EGFR, and cytokeratins (CKs) serves as key phenotypic biomarkers for precision oncology. We developed an AI-based image analysis tool that rapidly captures these transitions in cell assays, including specific protein biomarkers expressed on circulating tumor cells (CTCs).

Methods

We extended an ImageJ macro to enable rapid and reproducible extraction of biophysical parameters. The macro processes .lif, .nd2, and .czi file formats, using DAPI for nuclear segmentation and fluorophore-conjugated antibodies to delineate cytoplasmic boundaries. We evaluated automatic channel detection, intensity normalization, Otsu thresholding, and per-cell quantification of parameters such as surface area, circularity index (CI), and mean fluorescence intensity. Violin plots illustrated temporal variations in CI across A549 and MCF7 cells. Validation was conducted on CTCs isolated from cancer patient samples (n = 100) for PD-L1 and HER2 expression.

Results

The macro reduced image processing time from 7 minutes to 3 seconds per sample. A549 cells showed higher and more consistent CI values across all time points, while MCF7 cells demonstrated lower CI with greater variability, particularly at 24 and 72 hours. Quantitative measurements of PD-L1 and HER2 expression showed 100% concordance between the ImageJ macro and Zeiss software outputs, confirming analytical accuracy. CK18 intensity (~60–400) and PD-L1 (~20–50) levels measured by both platforms validated the macro’s ability to detect a wide range of marker expression in CTC subsets. CTCs exhibited higher CI values and greater morphological heterogeneity, consistent with their invasive phenotype.

Conclusions

We present an AI-driven macro that quantifies the biophysical characteristics of cancer cells, enabling precise phenotypic profiling, including circularity index, proliferation rates, and overexpression of biomarkers such as PD-L1 and HER2 in both cultured cell lines and patient-derived CTCs.