About PapSmearDB

PapSmearDB is a specialized medical image repository developed to support research in cervical cancer detection using modern computational techniques such as machine learning, deep learning, and medical image analysis. The platform provides access to curated Pap smear cytology images that can be utilized by researchers, clinicians, and students for developing automated diagnostic systems and computer-aided screening tools.

Cervical cancer remains one of the leading causes of cancer-related deaths among women worldwide. Early detection and timely medical intervention can significantly reduce mortality rates. One of the most effective screening methods used globally is the Papanicolaou test (Pap smear test). During this procedure, cervical cells are collected and examined under a microscope to identify abnormal cellular structures that may indicate precancerous or cancerous changes.

Pap Smear Cytology Classification

Pap smear images are commonly categorized according to the Bethesda System for Reporting Cervical Cytology, which is an internationally accepted standard used in cytopathology. The primary diagnostic categories include:

• Superficial –The outermost, most mature flat cells with small nuclei, indicating high estrogen.
• Intermediate –Middle-layer polygonal cells with abundant cytoplasm and slightly larger nuclei.
• Basal – Smaller, rounder immature cells found deeper in the epithelium.
• Parabasal – : The deepest, smallest stem cells attached to the basement membrane.
• LSIL – Low-grade lesion (mild dysplasia / CIN 1) often tied to temporary HPV infection, affecting superficial or intermediate cells.
• HSIL – High-grade lesion (moderate to severe dysplasia / CIN 2/3) with higher risk of progression.
• CIS – Carcinoma in situ, representing full-thickness severe nuclear atypia without invasion, now generally grouped under HSIL/CIN 3.

Purpose of the Dataset

The PapSmearDB dataset repository provides high-resolution microscopic images of cervical cytology slides to support the development and evaluation of automated diagnostic systems.

The dataset is useful for multiple research areas including:

• Cell classification and recognition
• Object detection of abnormal cervical cells
• Medical image segmentation
• Deep learning-based diagnostic systems
• Explainable artificial intelligence in healthcare

PapSmearDB promotes collaboration among researchers and supports the development of reliable computer-aided diagnostic tools for early detection and clinical decision-making.