Helmholtz Gemeinschaft

Search
Browse
Statistics
Feeds

Pan-Multiplex (Pan-M) dataset

Item Type:Dataset
Title:Pan-Multiplex (Pan-M) dataset
Creators Name:Rumberger, J.L.
Abstract:This dataset was constructed to train the Nimbus model for the publication "Automated classification of cellular expression in multiplexed imaging data with Nimbus". The dataset contains multiplexed images from different modalities, tissues and protein marker panels. It was constructed by a semi-automatic pipeline, where the cell types assigned by the authors of the original studies that published the data, where mapped back to their expected marker activity. In addition, for 3 FoVs of each dataset, 4 expert annotators proofread ~1.1M annotations which served as the gold-standard for assesing the algorithm. More details to the construction of the dataset can be found in the paper. The dataset consists of five subsets named codex_colon,mibi_breast,mibi_decidua,vectra_colon,vectra_pancreas, each in an individual folder. After unzipping, the data should be stored in the following folder structure to use the code provided for training and inference. To construct the binary segmentation maps used for training, you can use the code in segmentation_data_prep.py and simple_data_prep.py in the training repository.
Keywords:Multiplexed Imaging, Digital Pathology
Source:Hugging Face
Publisher:GitHub
Date:May 2025
External Fulltext:View full text on external repository or document server
Related to:

Repository Staff Only: item control page

Open Access
MDC Library