Presentation Information
[8a-P07-16]Image Reconstruction of Yeast Cells from Dual-Frequency Impedance Signals Using a CNN-Based Generative Framework
〇(D)Bela Hanief Abdurrahman1, Trisna Julian1, Tao Tang2, Yoichiroh Hosokawa1, Yaxiaer Yalikun2 (1.Nara Institute of Science and Technology, 2.Chongqing General Hospital, China)
Keywords:
microfluidics
Microfluidic impedance cytometry has emerged as an effective label-free modality for high-throughput analysis of individual particles. Despite its strengths, a key bottleneck remains: converting complex electrical signals into meaningful visual information, particularly under rapid flow conditions where traditional imaging methods often fail. To overcome this limitation, we introduce an AI-driven framework capable of reconstructing microscopic particle images directly from multi-frequency impedance measurements acquired in real time. The platform utilizes a coplanar three-electrode microfluidic device, structured on a Cr/Au-coated glass substrate. A PDMS microchannel (dimensions: 37 um width x 10 um height) is integrated to ensure narrow confinement, thereby minimizing particle overlap and promoting single-file transit across the sensing zone. Alternating current (AC) signals at two frequencies (1 MHz and 6 MHz) are applied to the central electrode, while voltage differentials are detected at the adjacent electrodes. Signal acquisition and demodulation are executed via a custom FPGA board (see Figure 1), and the resulting impedance signals are digitized and temporally aligned with high-speed video footage (captured at 4000 fps using a 63x inverted microscope). By aligning electrical and optical data streams, this synchronization process enables reliable image reconstruction of rapidly transiting particles. The elevated frame rate enhances event detection within brief acquisition intervals, supporting efficient high-throughput data collection. Ultimately, this strategy links real-time electrical signal measurement with image-based morphological interpretation, offering a robust alternative to conventional optical microscopy in flow-based cytometry systems.