APCEM-Enhanced Cervical Cytology Image Dataset
Description
A meticulously determined compilation of 4,253 cytology of cervical Pap smear images, representing multiple phases of epithelial abnormalities along with normal cellular conditions, makes up the APCEM-Enhanced Cervical Cytology Image dataset. The dataset is intended to assist studies in cytological image analysis, medical image modification, and cervical cancer screening. In order to enhance image quality while preserving cytological structure fidelity and diagnostic significance, all images have been manipulated using the approach recommended APCEM (Adaptive Parameterized Cytology Enhancement Model). Research Hypothesis Pap smear images visual quality and structural clarity can be significantly improved by carefully constructed traditional image enhancement techniques that adjust to cytological image properties without sacrificing diagnostic morphology. Improved interpretability and higher-quality inputs for subsequent analytical and computational studies in cervical cancer research are made possible by such upgrading. What the Data Shows There are 4,253 cervical cytology images in the collection that were obtained under standard laboratory imaging circumstances. It comprises six cytological broad categories that are clinically significant: ASCH (119 images), ASCUS (340 images), HSIL (1,300 images), LSIL (1,241 images), NILM (779 images), and SCC (474 images). Realistic variations in staining intensity, cellular density, overlap, background artifacts, and disease severity are displayed in the images, which mirror the features of actual cytology slides. Notable Features • Standardized magnitude of cytology sights after APCEM preprocessing. • Six cytological classes brought about with the Bethesda System. • Modern adaptive restructuring pipeline without optimization based on learning. • Medical imaging applications such as contrast enhancement and noise mitigation. • Preservation of cytoplasmic equipment, cellular structure, and the edges of cells. How to Interpret and Use the Data The APCEM classical adaptive image enhancement pipeline, which involves resolution normalization, luminance-based contrast enhancement in LAB color space, edge-preserving non-local means denoising, and rough masking for structural refinement, was employing to improve the images in the APCEM-Enhanced Cervical Cytology Image dataset. While avoiding artificial feature amplification, the augmentation optimizes cellular visibility and contrast-to-noise ratio. The dataset can be deployed as a standardized preprocessing resource for cervical cytology research, as well as for statistical image analysis, enhancement benchmarking, and visual clarity evaluation. Potential Applications • Evaluation of cervical cytology image quality and augmentation. • Comparing traditional methods for improving medical images. • Foundational preprocessing for cervical cancer screening research. • Academic studies in computer vision, medical imaging, and cytopathology.
Files
Steps to reproduce
Source of Cytology Images Cervical cytology Pap smear images were obtained from routine laboratory preparations reflecting real-world screening conditions. Slides were stained using the standard Papanicolaou (Pap) protocol to preserve nuclear and cytoplasmic morphology. Digital images were captured from representative microscopic fields containing epithelial cells. The dataset includes realistic variations in stain intensity, cytoplasmic transparency, cellular overlap, inflammatory background, mucus, and debris, consistent with routine cytology practice. Imaging Setup and Acquisition Slides were examined using light microscopes equipped with digital imaging systems. Images were acquired at standard diagnostic magnifications used in cytopathology. Microscope-mounted cameras and high-resolution imaging devices captured fine nuclear and cytoplasmic details. Images were stored in standard color formats prior to enhancement. Dataset Composition and Class Labels The dataset contains 4,253 cervical cytology images categorized into six clinically relevant classes used in Pap smear screening: • ASCH: 119 images • ASCUS: 340 images • HSIL: 1,300 images • LSIL: 1,241 images • NILM: 779 images • SCC: 474 images These classes represent clinically interpretable cytological outcomes observed during cervical cancer screening. Preprocessing and Image Enhancement All images were processed using the Adaptive Parameterized Cytology Enhancement Model (APCEM), a deterministic, non–learning-based enhancement pipeline tailored for cytology images. The APCEM workflow includes: (1) spatial normalization to a uniform resolution; (2) adaptive contrast enhancement using LAB color space and CLAHE on the luminance channel; (3) edge-preserving noise reduction using non-local means filtering; and (4) controlled sharpening using unsharp masking to enhance nuclear and cytoplasmic boundaries. All steps were implemented using OpenCV (Python) to ensure repeatable enhancement. Software Environment and Reproducibility The workflow was executed in Google Colab using Python, OpenCV, and NumPy to support open access and reproducibility. The dataset can be fully reproduced by applying the APCEM pipeline with fixed parameters to the original images. No deep learning or data-driven optimization was used, ensuring transparency, consistency, and interpretability.
Institutions
- VIT University - Chennai CampusTamil Nadu, Chennai