A Novel Dataset for Nuclei Segmentation in Melanoma Histopathology

Mark Schuiveling, Willeke A.M. Blokx, Gerben E. Breimer, Karijn P.M. Suijkerbuijk, Daniel Eek, Mitko Veta

Keywords: Nuclei segmentation, histopathology, melanoma, TILs

Abstract: The presence of tumor-infiltrating lymphocytes (TILs) in melanoma is associated with decreased recurrence of primary melanoma and increased survival in metastatic melanoma patients treated with immune checkpoint inhibition. Existing nuclei segmentation models performance is low due to the ability of melanocytes to mimic other cell types and due to existing melanoma specific models utilizing older, sub-optimal techniques. In addition, existing models do not provide tissue annotations necessary for determining the localization of TILs whereas this might also hold predictive value. To address this, we created a melanoma specific dataset with nuclei and tissue annotations. In this paper we describe the methodology used to create the dataset. In addition, we provide preliminary baseline benchmarks.

Submission Number: 52