A novel automatic segmentation and tracking method to measure cellular dielectrophoretic mobility from individual cell trajectories for high throughput assay
Introduction
Since Pohl's first report in the mid-20th century of cellular movement induced by the interaction between a non-uniform electric field and dipoles inside a cell, the behaviour, called dielectrophoresis (DEP), has been widely employed for many diverse practical applications such as manipulation of cells, characterisation of cells, detection and identification of cells, and drug and phenotype screening inside a microfluidic device [1], [2]. For example, the cellular movement caused by the DEP force has been used for flexible trapping and manipulation of single cells, high-throughput separation and trapping of cells, and precise control of a single cell [1], [2], [3], [4], [5], [6]. DEP has also been used to characterise cellular properties such as polarity, capacitance, conductance, and electrochemical activity [[7], [8]–9]. Moreover, the detection and identification of cancer cells, mycobacterium semegmatis single cells, and human embryonic stem cells [[10], [11]–12], and the screening of anticancer and antibiotic drugs and phenotypes [[13], [14]–15] have also been accomplished by DEP movement techniques. Success in those applications requires accurate movement of cells with a flexible direction inside a microfluidic device and precise tracking of the trajectories of cells when recording cell movement image sequences. To achieve the accurate movement of cells, many researchers have modified the fabricated structure of electrodes, which generate a non-uniform field that leads to the DEP force, in order to accurately and flexibly control the non-uniform electric field inside their microfluidic devices. Coplanar electrodes [2,7,9,16], interdigitated electrode arrays [[10], [11]–12,17], bipolar electrodes [4], a quadrupole electrode [5], and even an optically induced electrode [8] have been developed. These structure modifications in microfluidic devices resulted in significant progress and many achievements in these applications.
However, the critical issue of tracking the trajectories of cells has not been seriously considered even though that issue plays a pivotal role in the successful achievement of a high-throughput DEP experiments. For example, while the cells are moving under a DEP force, only a single cell is tracked and recorded in each time-lapse image because of the constrained field of view, where the area captured is around just one cell of interest. In this approach, cellular trajectories can be easily examined and tracked from the image stack because there is only one cell in each image. Nevertheless, a tremendous number of images are needed to obtain statistically reliable trajectory data, which are used for the characterisation of cellular properties. Moreover, to generalize a certain cell type property, at least more than 100 cells should be observed. That is, the tracking and recording method of a single cell in each time-lapse image leads to an extremely large cost [2]. To avoid this cost, one approach is to record multiple cell trajectories in time-lapse images that are segmented into individual cells from each time-lapse image, and examine single cell trajectories on each segmented image using image processing software such as ImageJ and Cell Profiler and simple, customised segmentation algorithms [3–5,8,11,12,17,18]. The cost can be dramatically reduced by examining single cell trajectories from the images with only one cell. However, the following issues still remain for the examination of cellular trajectories: the bright-field microscope images used for tracking cellular trajectories always include noise, which originates from the partially low and blurred contrast boundary between the transparent and non-spherical features of most cells, and the intensity variation caused by overlapping cells and electrode patterns on a substrate, which result in distortion of cell segmentation and poor tracking accuracy [19]. Moreover, while numerous cells are moving in a solution under a DEP force, collision and detachment events between adjacent cells frequently occur. These congested events should be considered in the continuous analysis of consecutive images in order to accurately track cellular trajectories [20]. These drawbacks cause only a few cells to be tracked in each captured image and become a barrier to simultaneously tracking a large number of parallel single cell trajectories in an automatic, robust, efficient, and accurate manner, resulting in a lack of accurate segmentation and tracking algorithms of a multitude of cells moving under a DEP force in a microfluidic device [4].
In this paper, we report a novel automatic segmentation and tracking method for multiple single cell trajectories induced by a DEP force acting on the cells inside a microfluidic device. For the successful demonstration of the novel method, we developed a cell segmentation and tracking algorithm that allows for cell identification and segmentation, linking of the segmented cells between consecutive frames, and the elimination of a periodic electrode patterns. Using this algorithm, several hundred MCF-7 cellular trajectories on an interdigitated electrode array were successfully segmented and tracked from the images in which the movements of cells induced by varying between a positive DEP (pDEP) force and a negative DEP (nDEP) force in the same environment were recorded. The segmentation and tracking algorithm was also verified by a comparison between the cross-over frequency (fco) of MCF-7 cells using the data from cellular trajectories with results reported in the previous work [21]. In addition, using the newly developed algorithm we tracked numerous cellular movements between traps, caused by varying between a pDEP and a nDEP force, until the viability of cells remained constant, resulting in a good statistical agreement between the repeated results. These results indicate that the developed algorithm is reliable for studying cellular DEP mobility using cell trajectories measured with the algorithm.
Section snippets
Imaging-based DEP-cell experiment overview
The important modules for the study of cellular dielectric properties utilising the movement of cells under a DEP force are the time-lapse microscopic technique and image-processing analysis. Fig. 1 illustrates the study workflow. First, the living cells were harvested from multiwell plates. Then, we observed the cellular DEP behaviour using a LabVIEW-based automated DEP system that enabled us to simultaneously modulate microscopic image acquisition, AC electric frequency, and AC electric
Results for cell segmentation on a DEP chip
To characterise cellular DEP behaviour, the primary goal is to accurately identify and count the number of MCF-7 cancer cells from time-lapse optical images taken of the fabricated DEP chip. The cell segmentation must account for the dynamic change in cell image intensity caused by interferences from the background pattern as cells move on the chip, which consisted of periodic electrodes with patterned striped and circular regions. For example, Fig. 3A shows the cells isolated by different DEP
Discussion
Overall, the cell tracking accuracy and cell count capability reached at least 90% of the total number of cells with the algorithm developed in this work (Fig. 4C and D). According to our comparison study with a commercialized software method and a deep neural networks-based method, the developed method in this paper was superior for examining the mobility from individual cellular trajectories of an ensemble of cells influenced by a DEP force in the same environment as a microfluidic device
Conclusion
We developed a novel automatic segmentation and tracking method for analysing the cellular trajectories of an ensemble of cells influenced by a DEP force in the same environment as inside a microfluidic device. The tracking accuracy and count capability of the cell ensembles was greater than 90% over the entire cell population after the method was performed for the examination of the cellular trajectories recorded in time-lapse images. We also analysed cell trapping efficiency and occupancy on
Declaration of Competing Interest
All authors declare that no conflicts of interest exist.
Acknowledgements
This research was supported in part by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2017R1A2B2002076 and NRF-2019R1F1A1058971), Republic of Korea, in part by the Global Frontier Program, through the Global Frontier Hybrid Interface Materials (GFHIM) of the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2013M3A6B1078872) and in part by "Leaders in
References (38)
- et al.
Determination of cell membrane capacitance and conductance via optically induced electrokinetics
Biophys. J.
(2017) - et al.
MobilityAnalyser: A novel approach for automatic quantification of cell mobility on periodic patterned substrates using brightfield microscopy images
Comput. Meth. Programs Biomed.
(2018) Review article—dielectrophoresis: status of the theory, technology, and applications
Biomicrofluidics
(2010)- et al.
Rapid dielectrophoretic characterization of single cells using the dielectrophoretic spring
Lab Chip
(2013) - et al.
Comprehensive analysis of human cells motion under an irrotational AC electric field in an electro-microfluidic chip
PLOS ONE
(2014) - et al.
High-throughput separation, trapping, and manipulation of single cells and particles by combined dielectrophoresis at a bipolar electrode array
Anal. Chem.
(2018) - et al.
Flexible trapping and manipulation of single cells on a chip by modulating phases and amplitudes of electrical signals applied onto microelectrodes
Anal. Chem.
(2019) - et al.
Tunnel dielectrophoresis for tunable, single-stream cell focusing in physiological buffers in high-speed microfluidic flows
Small
(2016) - et al.
Multiparameter cell-tracking intrinsic cytometry for single-cell characterization
Lab on a Chip
(2018) - et al.
Microfluidic dielectrophoresis illuminates the relationship between microbial cell envelope polarizability and electrochemical activity
Scie. Adv.
(2019)
Dielectrophoresis based discrimination of human embryonic stem cells from differentiating derivatives
Biomicrofluidics
Label-free electric monitoring of human cancer cells as a potential diagnostic tool
Anal. Chem.
Dielectrophoresis as a single cell characterization method for bacteria
Biomed. Phys. Eng. Express.
Dielectrophoretic assay of bacterial resistance to antibiotics
Phys. Med. Biol.
Dielectrophoresis-based cellular microarray chip for anticancer drug screening in perfusion microenvironments
Lab Chip
Dielectrophoretic microfluidic chip enables single-cell measurements for multidrug resistance in heterogeneous acute myeloid leukemia patient samples
Anal. Chem.
Quantitative characterization for dielectrophoretic behavior of biological cells using optical tweezers
Appl. Phys. Lett.
Real-time analysis of cellular response to small-molecule drugs within a microfluidic dielectrophoresis device
Anal. Chem.
Movement of giant lipid vesicles induced by millimeter wave radiation change when they contain magnetic nanoparticles
Drug Deliv. Transl. Res.
Cited by (4)
Cellular subpopulations identified using an ensemble average of multiple dielectrophoresis measurements
2024, Computers in Biology and MedicineMachine Vision Based Method for Measuring Single-Cell Biophysical Properties Using Dielectrophoresis Mobility
2021, 21st International Conference on Solid-State Sensors, Actuators and Microsystems, TRANSDUCERS 2021A particle-capturing and -separating, sheathless chip in series with microfilters and planar interdigital electrodes
2021, Journal of Physics D: Applied Physics
- 1
These authors contributed equally to this work.