Augmentation-Based Deep Learning for Identification of Circulating Tumor Cells

TL;DR

CTC identification in liquid biopsy is hampered by rarity and heterogeneity. The authors develop a DL pipeline that learns from bright-field DEPArray single-cell images and uses DAPI-channel augmentation during training, while testing relies only on BF images. Across architectures, ResNet50 yields the best F1-score of $0.798$, with augmentation and DAPI contributing to generalization. The work shows that fluorescence-independent CTC discrimination is achievable and supports streamlined, automated workflows in clinical liquid biopsy.

Abstract

Circulating tumor cells (CTCs) are crucial biomarkers in liquid biopsy, offering a noninvasive tool for cancer patient management. However, their identification remains particularly challenging due to their limited number and heterogeneity. Labeling samples for contrast limits the generalization of fluorescence-based methods across different hospital datasets. Analyzing single-cell images enables detailed assessment of cell morphology, subcellular structures, and phenotypic variations, often hidden in clustered images. Developing a method based on bright-field single-cell analysis could overcome these limitations. CTCs can be isolated using an unbiased workflow combining Parsortix technology, which selects cells based on size and deformability, with DEPArray technology, enabling precise visualization and selection of single cells. Traditionally, DEPArray-acquired digital images are manually analyzed, making the process time-consuming and prone to variability. In this study, we present a Deep Learning-based classification pipeline designed to distinguish CTCs from leukocytes in blood samples, aimed to enhance diagnostic accuracy and optimize clinical workflows. Our approach employs images from the bright-field channel acquired through DEPArray technology leveraging a ResNet-based CNN. To improve model generalization, we applied three types of data augmentation techniques and incorporated fluorescence (DAPI) channel images into the training phase, allowing the network to learn additional CTC-specific features. Notably, only bright-field images have been used for testing, ensuring the model's ability to identify CTCs without relying on fluorescence markers. The proposed model achieved an F1-score of 0.798, demonstrating its capability to distinguish CTCs from leukocytes. These findings highlight the potential of DL in refining CTC analysis and advancing liquid biopsy applications.

Figures (4)

• Figure 1: Different types of images produced by DEParray.
• Figure 2: Representative images of CTC and non-CTC cells in both DAPI and BF microscopy.
• Figure 3: Workflow of the proposed method.A. Liquid biopsy is a blood test that detects cancer cells or tumor DNA, avoiding invasive procedures. B. The DEPArray™ system uses electric fields to isolate and select single cells, like CTCs. C. Training image pre-processing consist of BF DEPArray images whose variability is increased through the use of augmentation operations, and fluorescent-field images of DAPI-labeled cells. D. The output of the CNN is the classification of images into CTC and non-CTC. E. BF images are used in the validation/test phase to identify CTCs.
• Figure 4: Statistical results of Mann-Whitney test a) Statistical results with $p-value<0.05$ demonstrate a statistically significant difference between the two groups considered (BF and BF with DAPI). b) The data do not follow a normal distribution. c) Mann-Whitney test. d) Homogeneity test.