Circulating Tumor Cell Identification Based on Deep Learning

Abstract

As a major reason for tumor metastasis, circulating tumor cell (CTC) is one of the critical biomarkers for cancer diagnosis and prognosis. On the one hand, CTC count is closely related to the prognosis of tumor patients; on the other hand, as a simple blood test with the advantages of safety, low cost and repeatability, CTC test has an important reference value in determining clinical results and studying the mechanism of drug resistance. However, the determination of CTC usually requires a big effort from pathologist and is also error-prone due to inexperience and fatigue. In this study, we developed a novel convolutional neural network (CNN) method to automatically detect CTCs in patients’ peripheral blood based on immunofluorescence in situ hybridization (imFISH) images. We collected the peripheral blood of 776 patients from Chifeng Municipal Hospital in China, and then used Cyttel to delete leukocytes and enrich CTCs. CTCs were identified by imFISH with CD45+, DAPI+ immunofluorescence staining and chromosome 8 centromeric probe (CEP8+). The sensitivity and specificity based on traditional CNN prediction were 95.3% and 91.7% respectively, and the sensitivity and specificity based on transfer learning were 97.2% and 94.0% respectively. The traditional CNN model and transfer learning method introduced in this paper can detect CTCs with high sensitivity, which has a certain clinical reference value for judging prognosis and diagnosing metastasis.

Image-based cell sorting using focused travelling surface acoustic waves

“…Different neural networks models have been developed to classify cells. Dendritic cells have been counted reliably via image processing algorithms (CasDC) for cancer treatment monitoring, 14 macrophages, 62 B and T lymphocytes, 27 stem cells 63 and circulating tumor cells 64 have been also identified based on label-free image-based analysis. DNN-based sorting has been demonstrated in our first paper 15 and more recently improved and extended to single cells from nervous tissue.…”

Machine Learning-Driven Multiobjective Optimization: An Opportunity of Microfluidic Platforms Applied in Cancer Research

“…This innovative combination of two techniques, intelligent microfluidics, not only implements the dominant conveniences of microfluidics, but also uses the strength of ML to help microfluidic platforms live up to their enormous potential. Microfluidic platforms applied in cancer research have two major applications: CTC isolation [ 3 , 8 , 9 ] and tumor modeling to mimic metastatic microenvironments.…”

“…Label-free-based strategies stand for CTC separation by biophysical properties other than the distinctive biomarkers on the cancer cells’ surface and the nucleic acid inside the cytoplasm [ 3 , 17 , 18 , 19 ]. CTCs are isolated from blood cells by their biophysical properties, such as cell morphology, buoyant density, electric charge, and deformability (see Figure 2 b), to facilitate subsequent downstream analysis in cancer metastasis [ 18 , 19 , 20 , 21 ].…”

“…In contrast to traditional statistic-based data analysis, ML features the dynamic advancement of predication aided by computers that can deal with large amounts of data, decreasing human intervention and workload. As mentioned above, over 50% of the applications of intelligent microfluidics in biotechnology include cell classification, cell screening, cell sorting, the identification of cellular pathology, and the measurement of cellular chemistry [ 3 , 18 , 21 ]. ML can also be used in the analysis, design, and manipulation of continuous or separate fluids in micro-constructs in order to optimize microfluidic platforms [ 14 , 17 , 31 , 32 , 33 ].…”

“…Circulating tumor cells (CTCs) can serve as a primary indicator of metastasis [ 2 ]. The isolation, identification, and characterization of CTCs are meaningful for metastasis research [ 3 ]. Traditional approaches to cancer metastasis are limited in physiological relevance in microenvironments (particularity with invasion assays based on Petri dishes), lack quantification accuracy with trans-well-based transmigration assays, and have challenges in spatiotemporal control in animal experiments.…”