Mel-DEPTHS: A Benchmark Melanoma Histopathology Dataset
This retrospective dataset comprises 50 whole-slide images (WSIs) from patients diagnosed with melanoma at the Department of Pathology, Istanbul University–Cerrahpaşa, Cerrahpaşa Faculty of Medicine Hospital, between 2018 and 2022. Each patient contributes a single WSI. All slides passed a two-stage quality control by a dermatopathologist, who has >20 years of experience, ensuring appropriate staining, tissue integrity, and histologic adequacy.
Selected glass slides were scanned with an APOCHROMAT objective (20×/0.75 NA). Combined with the camera optics, this yields an effective 40× magnification (0.25 µm/pixel). After scanning, each WSI was reviewed again for residual artifacts. Minor variations (e.g., slight color shifts, small focus inconsistencies) reflecting real-world practice were accepted. Final images were saved as SVS files , a multi-tiled, TIFF-based file format.
The dataset is most strongly represented in 2019 with four patients, 2021 with 20 patients, followed by 2020 and 2018, each with 12 patients.There are 2 cases in 2022, the project’s starting year. Tumor stages are distributed according to the PT classification: six cases at PT1, eight at PT2, fifteen at PT3, and twenty-one at PT4.
Supervisor
- Prof. Songül Varlı
Department of Computer Engineering, Yıldız Technical University
Principal Investigator
- PhD. Yasemin Topuz
Department of Computer Engineering, Yıldız Technical University
Researchers
- Prof. Övgü Aydın Ülgen
Department of Pathology, Istanbul University-Cerrahpasa
- Asst. Prof. A. Mine Önenerk Men
Department of Pathology, Istanbul University-Cerrahpasa
- Asst. Prof. Nurullah Calik
Department of Biomedical Engineering, Istanbul Medeniyet University
Fellowship Students
- M. Taha Gökcan
- Serdar Yıldız
- İpek Sertbudak
- Serhat Kaymaz
- Özge Gökbaşı
- Nil Urgancı
To obtain the dataset: Dataset Link
Directory Structure
Acknowledgement
This research was funded by TUSEB (Health Institutes of Türkiye) Group B R&D Projects Project Number: 24348.
Ethical approval was obtained from the Non-Invasive Clinical Research Ethics Committee of Istanbul University–Cerrahpaşa (Decision No: 2022/117).
License
Creative Commons 4.0 BY (Attribution) NC-ND (Non-Commercial, No Derivatives) license.
https://creativecommons.org/licenses/by-nc-nd/4.0/
Reference
Topuz, Y., Gökcan, M. T., Men, A., Yıldız, S., Sertbudak, İ., Kaymaz, S., … & Varlı, S. (2026). Mel-DEPTHS: a benchmark dataset for epidermis and tumor segmentation for melanoma staging. Medical & Biological Engineering & Computing, 1-18.