Supporting Data for "Advancing Biophysical Cytometry with Deep-learning (ABCD) for New-Generation Cell-based Diagnostics"

<p>There are in total of 7 types of lung cancer cells (i.e. H69, H358, H520, H526, H1975, H2170 and HCC827). All the data were collected on 7 days using <a href="https://doi.org/10.1002/jbio.201800479" rel="nofollow">multi-ATOM setup</a>, giving 3 batches per cell lines. Both single-cell brightfield and quantitative phase images (QPI) were collected.</p><p>For training and testing the beGAN model, the data were separated into "Train", "Valid" and "Test" set. They are subsampled and contain 500 cells respectively as a demonstration in this repository (Folder <code>Dataset</code>). Data was uploaded in <code>.mat</code> format with brightfield images in <code>_BF.mat</code> and QPI in <code>_QPI.mat</code>. The images are stored in format of <code>ImageHeight * ImageWidth * NoOfCells</code> with a field of view of 45μm.</p>