This dataset contains Human Promyelocytic Leukemia (HL60) cancer cell images with magnification factor 40x by hemocytometer (HEM40X) for cell counting (CC). It has total 468 raw hemocytometer images acquired on 40x light-microscope objectives. Also, annotation information by domain experts shared as total 6890 cells.
The dataset consist of three complementary components, namely; Image Sets, Ground Truth Cell Annotations and Counting Area (ROI) Boundary Annotations.
Images are acquired in different sessions by Motic B3-Series 2.0 Megapixels Moticam 2000 camera attached to the microscope. Moreover, the dataset contains 468 Red-Green-Blue (RGB) images in 1200x1600 pixel resolution. In particular, imperfect visualization, cell shape deformations, varying lighting, clumped cells and impurities are some of the real conditions observed in the dataset.
Randomly chosen 4 sample images given below from our gathered unique dataset those have possible adverse conditions, including, deformed cell shape, images with different brightness, single cell image with unequal brightness, images having clumped cells and impurities in cell suspension.
Ground truth cell coordinates for each image are annotated by three experts by labelling 6890 cells in total. Experts annotate all cell locations as ‘Positive’ and non-cell locations as ‘Negative’. Randomly chosen sample croped images of cell and non-cell are given below from our dataset.
For each image, the counting area boundary annotations are defined by employing the boundaries and they are manually annotated by the experts. Ultimately, each of the annotations are served separate file as the same format cell annotations. Each of the file has four entities which define the top, bottom, right and left boundaries of the counting area respectively. More precisely, the cells intersecting left and top sides of the middle of the triple lines are counted (i.e. counting area) yet the cells on the right and bottom ones are not counted in the process.
For the below figure, Left is considered one of the sample image from our proposed dataset with corresponding ground truth annotations as matrix form. Cell locations are marked by green bounding boxes and four Counting Area (ROI) Boundary Annotations drawn by blue bounding boxes. The inside of area covered by these four squares is called the Counting Area (ROI).
If you use this dataset, please cite the following paper:
Özkan, A., İşgör, S. B., Şengül, G., & İşgör, Y. G. (2018). Computer vision based automated cell counting pipeline: a case study for HL60 cancer cell on hemocytometer. Biomedical Research, 29(14), 2956-2962. [pdf] [BibTex]