PABPS: Paired Brightfield and Papanicolaou-Stained Sperm Microscopy Dataset

Overview PABPS (Paired Brightfield and Papanicolaou-Stained Sperm) contains 499 pairs of microscopy images of ethanol-fixed human spermatozoa from five healthy volunteers (donors D101–D105). Each pair consists of an unstained brightfield (BF) image and an image of the same field of view after Papanicolaou (Pap) staining. The data accompany the manuscript listed under Related works. Contents The record holds one archive, PABPS-v1.0.zip (13.86 GB), and its checksum file PABPS-v1.0.zip.sha256. - 998 TIFF images (499 BF, 499 Pap), 3072 × 2048 px, 8-bit RGB, 0.109 µm/px- Leica XML metadata for each image- file_manifest_v4.csv (one row per pair), donor_summary.csv, SHA256SUMS.txt- README.md (directory layout and column definitions)- examples/pabps_dataset.py (PyTorch data-loading example) Data collection Semen was washed, spread on a grid positioning slide, air-dried and fixed in 95% ethanol. The unstained BF image of each field was acquired first. The slide was then Pap-stained and the same grid coordinates were imaged again. Images were acquired on a Leica DMIL inverted microscope with an HC PL FLUOTAR L 40×/0.60 DRY objective and a Leica K3C camera (LAS X 3.7.6). Staining used the Modified Papanicolaou Staining Solution kit from Cariad Medical Technology (Zhuhai, China; lot 0230). Registration The two images of a pair are unregistered originals (median offset 3.0 µm). Users should register each pair before pixel-wise training or evaluation. The offset between paired images, image sharpness and stain colour are reported in the Technical Validation section of the manuscript. Pairs per donor D101: 109; D102: 113; D103: 106; D104: 98; D105: 73 (total 499). Use Intended for BF-to-Pap image translation, image registration and stain normalisation. Donor, imaging session and stain batch are confounded, so split by donor (for example, leave-one-donor-out). Not intended for clinical diagnosis. Ethics and de-identification Approved by the Medical Ethics Committee of the Women and Children's Hospital of Chongqing Medical University (approval (2026) LunShen (Ke) No. 088; 20 July 2026). All participants gave written informed consent, including consent to public release of de-identified images. Names, hospital identifiers and dates of birth appear in neither file names nor embedded metadata. Donor-level data are limited to donor code, age, first BF imaging date, grid zones and pair counts (donor_summary.csv). Integrity check Verify the downloaded archive against its checksum file, in the folder containing both files: shasum -a 256 -c PABPS-v1.0.zip.sha256 (macOS) or sha256sum -c PABPS-v1.0.zip.sha256 (Linux). After extraction, verify the individual files in the extracted folder: shasum -a 256 -c SHA256SUMS.txt (macOS) or sha256sum -c SHA256SUMS.txt (Linux). FundingThis research was supported by private financial funding from Jiang Wang. No grant or award numbers apply.AcknowledgementsThe authors express their deepest gratitude to Jiang Wang, whose foundational support and enduring trust made this research possible and sustained it to completion. The authors are immensely grateful to Wei Sun for his tireless dedication, vital facilitation, and indispensable support throughout the project. The authors also thank Heng Meng for facilitating the collaboration, and Tingting Lin's Laboratory for providing the workspace and experimental equipment. License and citation CC BY 4.0. Please cite this record and the accompanying manuscript.