An annotated fluorescence image dataset for training nuclear segmentation methods

Florian Kromp,Klaus Beiske,Tamara Weiss,Sabine Taschner-Mandl,Maria Berneder,Eva Bozsaky,Peter F Ambros,Daria Lazic,Inge M Ambros,Magdalena Ambros,Fikret Rifatbegovic,Allan Hanbury,Lukas Fischer,Wolfgang Dörr

doi:10.1038/s41597-020-00608-w

Abstract

Fully-automated nuclear image segmentation is the prerequisite to ensure statistically significant, quantitative analyses of tissue preparations,applied in digital pathology or quantitative microscopy. The design of segmentation methods that work independently of the tissue type or preparation is complex, due to variations in nuclear morphology, staining intensity, cell density and nuclei aggregations. Machine learning-based segmentation methods can overcome these challenges, however high quality expert-annotated images are required for training. Currently, the limited number of annotated fluorescence image datasets publicly available do not cover a broad range of tissues and preparations. We present a comprehensive, annotated dataset including tightly aggregated nuclei of multiple tissues for the training of machine learning-based nuclear segmentation algorithms. The proposed dataset covers sample preparation methods frequently used in quantitative immunofluorescence microscopy. We demonstrate the heterogeneity of the dataset with respect to multiple parameters such as magnification, modality, signal-to-noise ratio and diagnosis. Based on a suggested split into training and test sets and additional single-nuclei expert annotations, machine learning-based image segmentation methods can be trained and evaluated.

Highlights

Bioimage analysis is of increasing importance in multiple domains including digital pathology, computational pathology, systems pathology or quantitative microscopy[1,2,3,4,5,6]
Publicly available databases and platforms allow the access to image datasets and annotations enabling the research community to develop sophisticated algorithms for complex bioimage analysis
Digital pathology relies on tissue sections as a basis for diagnosing disease type and grade or stage[7,8]

Summary

Background & Summary

Bioimage analysis is of increasing importance in multiple domains including digital pathology, computational pathology, systems pathology or quantitative microscopy[1,2,3,4,5,6]. Remaining aggregations of nuclei lead, in the worst case, to their exclusion from the downstream analysis potentially causing a biologically significant bias To overcome these drawbacks, novel deep learning-based image segmentation approaches are currently developed in many labs worldwide. We hereby present an expert-annotated comprehensive dataset[18] that can be used to train machine learning-based nuclear segmentation algorithms. The final annotated dataset, forming the ground truth dataset, was split into a training set and a test set to be used for machine learning-based image segmentation architectures. The proposed expert-annotated dataset presents a heterogeneous, real-world dataset consisting of fluorescence images of nuclei of commonly used tissue preparations showing varying imaging conditions, sampled using different magnifications and modalities. The dataset can be used to train and evaluate machine learning-based nuclear image segmentation architectures, thereby challenging their ability to segment each instance of partially highly agglomerated nuclei

Methods

Code availability