Blood Cell Detection Dataset
| Modality | Data Format | Publisher | Licence |
| Light Microscope | PNG | Abdüssamet Aslan | MIT |
What is a Peripheral Blood Smear?
A peripheral blood smear is a cornerstone in the field of hematology, providing crucial insights into various blood disorders and infections. It involves smearing a thin layer of blood on a glass microscope slide, followed by specialized staining to facilitate the microscopic examination of different blood cells. These smears are invaluable for investigating hematological conditions, including but not limited to anemia, leukopenia, and thrombocytopenia. Furthermore, they are routinely used for identifying blood-borne parasites, such as those causing malaria and filariasis.
In an era where automated blood analyzers are becoming increasingly prevalent, the art of examining a peripheral blood smear retains its relevance as a powerful diagnostic tool. It provides a snapshot of the bone marrow’s functional status—the body’s “factory” for all blood elements. Reviewing a peripheral blood smear often complements other clinical data and, in some cases, may suffice to confirm a diagnosis. As such, the Blood Cell Detection Dataset is an invaluable asset for training machine learning models and augmenting medical research in the domain of hematology.
About the Blood Cell Detection Dataset
The Blood Cell Detection Dataset is a robust resource for researchers, medical professionals, and data scientists interested in the automated detection and analysis of blood cells. It comprises high-quality images captured from peripheral blood smears via a light microscope, achieving high magnification and resolution. The dataset consists of 100 annotated images focusing on red blood cells (RBCs) and white blood cells (WBCs). Specifically, it features 2237 annotated RBCs and 103 annotated WBCs.
Each image in the dataset is a full-color RGB image with dimensions of 256×256 pixels, allowing for a detailed examination of individual cells. The images are stored in PNG format, ensuring a lossless representation of the original microscope captures. Accompanying these image files is an ‘annotations.csv’ file that provides essential meta-information, such as the locations and labels of the annotated blood cells. The entire dataset is relatively lightweight, with a total file size of approximately 14MB. Contributions to improve or extend the dataset are openly encouraged, and all pull requests (PRs) are welcome.