Cervical Cancer Cell Image Database: Multi-focus Cytology Dataset (CCCID)

This dataset contains cervical cytology cell images classified into four diagnostic categories according to the Bethesda System: NILM (Negative for Intraepithelial Lesion or Malignancy), LSIL (Low-grade Squamous Intraepithelial Lesion), HSIL (High-grade Squamous Intraepithelial Lesion), and SCC_Others (Squamous Cell Carcinoma and other malignant lesions, including adenocarcinoma). The images were obtained from Papanicolaou-stained cervical cytology specimens and digitized using a whole-slide scanner (20× objective equivalent). Each cell image is provided as a nucleus-centered 11-layer multi-focus image stack with a focus interval of 1 μm. This dataset was developed to support AI-assisted cytology research, image standardization, annotation organization, open data sharing via Zenodo, and future research in cervical cytology image analysis. The database was originally developed as part of the JST-CREST research project "Construction of an Innovative Cancer Diagnosis Support System Using 3D Image Recognition AI" and organized by the Cervical Cancer Cell Image Database Project (hereafter referred to as the CCCID-Project).