Human somatic label-free bright-field cell images
This cell images dataset is collected using an ultrafast imaging system known as asymmetric-detection time-stretch optical microscopy (ATOM) for training and evaluation. This novel imaging approach can achieve label-free and high-contrast flow imaging with good cellular resolution images at a very high speed. Each acquired image belongs to one of the four classes: THP1, MCF7, MB231 and PBMC. Our objective is to complement it with a high-speed classification system that is sensitive to the representations of different cell types, and meanwhile should ignore the noise caused by varying experimental settings. We have at our disposal two collections of images captured separately at different instances. Dataset_1 contains over 8,000 samples of three types of cells (MCF7, PBMC, and THP1), while the other one Dataset_2 is significantly larger, with over 900,000 cell images of MB231, MCF7, and THP1 cells.
This dataset is used for Paper: Large-scale Multi-class Image-based Cell Classification with Deep Learning
- A smaller cell dataset for cell classification. Three types of cells are included, namely, THP1, MCF7, PBMC dataset1.zip (144.03 MB)
- Large-scale cell dataset for cell classification. Three types of cells are included, namely, THP1, MCF7, MB231 dataset2.zip (1.41 GB)