CPSMI2025
Description
This dataset is related to a Data in Brief paper named "CPSMI2025: A Curated Dataset of Conventional Pap Smear Microscopy Images for Deep Learning-Based Cervical Cancer Screening". It comprises 2,169 images of conventional Pap smears obtained using a low-cost, open-design microscope. The images are organized into nine subcategories, which are further grouped into four major categories. They were captured from more than 350 different slides, each previously reviewed and interpreted by both a pathologist and a cytotechnologist.
Files
Steps to reproduce
The images were acquired using a custom-built, replicable low-cost microscope integrated with a Raspberry Pi HQ camera featuring a 12.3-megapixel Sony IMX477 sensor. All images were captured at the sensor's native resolution using an automatic exposure program. Subsequently, they were downsampled to one-quarter of their original dimensions using Python's Pillow library, employing the LANCZOS filter for high-quality resampling.
Institutions
- Universidad Nacional Autonoma de Mexico
- Colegio De La Frontera Sur