An electrical imaging chip has tracked the movement of cells without fluorescent labels, from individual MDA-MB-231 breast-cancer cells to compact cell clusters and millimetre-scale tissues, in a laboratory study. The system was designed to follow both single-cell paths and collective spreading across a much larger area than the optical comparison.
The platform uses a 1024-by-1024 sensing array on a 10-micrometre grid. Its active area is 10.24 by 10.24 millimetres, or 104 square millimetres, and it was fabricated in a 180-nanometre CMOS process. Capacitance readings from the chip were converted into images that could be segmented and followed over time.
From single cells to a wider field
The researchers worked with MCF-10A mammary epithelial cells and MDA-MB-231 triple-negative breast-cancer cells, including fluorescent reporter variants. To test cell detection, they compared capacitance maps with fluorescence images and trained CellPose cyto3 on about 1,000 manually annotated cell masks. The two approaches generally placed cells in similar positions, but fluorescence produced smaller cell boundaries. A measure of cell shape called eccentricity remained similarly correlated without a systematic bias.
At a sparse seeding density of about 5%, MDA-MB-231 cells were imaged for 20 hours. The capacitance signal kept cells visible through division, and the tracking software reconstructed trajectories that matched the overlaid capacitance images, including new tracks for daughter cells. Its pipeline estimated movement between frames, matched detections across time and filled gaps of up to eight frames, corresponding to roughly 160 minutes.
The larger benefit appeared when the researchers compared migration across setups. Optical-flow migration metrics were comparable on tissue-culture plastic and on the CMOS-MEA, but the optical field of view measured 1.6 by 1.3 millimetres, or 2.08 square millimetres, versus 156 square millimetres for the CMOS-MEA. The comparison used different imaging setups and did not report replicate-level uncertainty, so it supports a large-area capability more clearly than a claim of superior measurement accuracy.
Collective movement comes into view
The chip also followed collective migration in MDA-MB-231 spheroids, prepared with a target of about 7,200 cells per spheroid. The spheroids began at about 350 micrometres in diameter and spread for 50 hours. Spreading slowed over time, with the reported effective radius following an approximate power-law relation with an exponent of about 0.4. The fastest outward movement occurred in the outermost 300 micrometres, while nearby velocities were coordinated over a characteristic length of 100 micrometres.
To probe tissue shape, the researchers used mutual capacitance and changed the size of the sensor's virtual pixels, groups of readings treated as one measurement. Simulations estimated a critical detection height of 4 micrometres at the smallest setting and 60 micrometres at the largest tested setting. In practical terms, larger groupings were more sensitive to elevated features but gave up lateral detail. The experimental and simulated patterns agreed qualitatively, so the height information remains a simulated interpretation rather than a direct three-dimensional measurement.
At the largest scale, MCF-10A honeycomb tissues made from a target of 140,000 cells were 3 millimetres across. Their projected area stayed near 2.5 square millimetres for the first eight hours, then increased at about 0.3 square millimetres per hour. The coordinated peripheral region widened from about 130 to 500 micrometres. The tissue spread through interior openings and along its outside edge, with the interior front moving at 39.96 micrometres per hour compared with 28.91 micrometres per hour at the exterior boundary.
What the system can and cannot show
The authors describe the platform as scalable, but the evidence remains a laboratory proof of concept in selected cell lines and engineered tissues. Its spatial resolution is coarser than conventional optical microscopy, and the virtual-pixel modes trade lateral resolution and readout speed for depth sensitivity. The reported tracking and spreading results are descriptive, with no formal tracking accuracy estimate or replicate-level uncertainty reported in the supplied analysis.
The paper states that its phase-contrast, fluorescence and capacitance measurements, along with analysis data, are available on Zenodo, with supporting Python scripts on GitHub. Funding came from Brown University awards, National Science Foundation grant 2027108 and National Institutes of Health grant R01GM140108. The authors report no competing financial interests.
Paper data and sources
Original title: Label-free all-electrical tracking of individual and collective cell migration on a megapixel CMOS capacitance sensor
Authors: Jeong H, Joshi P, Hu Y et al.
Journal/Repository: Lab on a chip
Status: Peer-reviewed
First online: 2026-10-19
DOI: Not available
Original paper · Full text