Electrical Capacitance Tomography of Cell Cultures on a CMOS Microelectrode Array.

Area of Science:

• Biomedical Engineering
• Imaging Technology
• Computational Biology

Background:

• Electrical Capacitance Tomography (ECT) enables non-invasive internal volume imaging via boundary capacitance measurements.
• Existing ECT systems often lack the resolution required for microscopic biological structures.
• Advanced imaging techniques are crucial for understanding complex cell cultures.

Purpose of the Study:

• To develop a microscale capacitance tomography system with high spatial resolution.
• To create a deep learning model for accurate 3-D reconstruction of cell cultures.
• To demonstrate a low-cost, label-free 3-D imaging tool for biological samples.

Main Methods:

• Implementation of a microscale capacitance tomography system utilizing an active CMOS microelectrode array with 10-micron resolution.
• Development and training of a deep learning model for 3-D volume reconstruction from capacitance data.
• Application of a multi-objective loss function (pixel-wise, distribution-based, region-based) to enhance reconstruction accuracy.

Main Results:

• The microscale capacitance tomography system achieved a spatial resolution of 10 microns.
• The deep learning model with a multi-objective loss function improved reconstruction accuracy by 3.2% compared to pixel-wise loss alone.
• The model demonstrated an average 4.6% improvement over baseline computational methods on evaluated datasets.
• Successful 3-D imaging of bacterial biofilms, resolving microscopic spatial features.

Conclusions:

• Microscale capacitance tomography offers a high-resolution, label-free 3-D imaging solution for cell cultures.
• The developed deep learning approach significantly enhances reconstruction accuracy in microscale ECT.
• This technology presents a promising low-cost, low-power tool for biological sample analysis.