Evaluation of cell segmentation methods without reference segmentations

Cell segmentation is a cornerstone of many bioimage informatics studies. Inaccurate segmentation introduces computational error in downstream cellular analysis. Evaluating the segmentation results is thus a necessary step for developing the segmentation methods as well as choosing the most appropriate one for a certain kind of tissue or image. The evaluation process has typically involved comparison of segmentations to those generated by humans, which can be expensive and subject to unknown bias. We present here an approach that seeks to evaluate cell segmentation methods without relying upon comparison to results from humans. For this, we defined a number of segmentation quality metrics that can be applied to multichannel fluorescence images. We calculated these metrics for 11 previously-described segmentation methods applied to datasets from 5 multiplexed microscope modalities covering 5 tissues. Using principal component analysis to combine the metrics we defined an overall cell segmentation quality score and ranked the segmentation methods. A Reproducible Research Archive containing all data and code will be made available upon publication at http://hubmap.scs.cmu.edu.

Deep Collection of Quantitative Nuclear Division Dynamics Data in RNAi-treated Caenorhabditis elegans Embryos

SUMMARYRecent advances in bioimage informatics techniques have yielded quantitative data on multicellular dynamics from microscopy images of animal development. Several such data collections have been created for Caenorhabditis elegans embryos under various gene silencing conditions. However, because of the limited depth of the datasets, it is impractical to apply standard statistical methods to these collections. Here, we created a deep collection of quantitative data on nuclear division dynamics during the first three rounds of cell division in C. elegans embryos, in which 263 essential embryonic genes were silenced individually by RNA-mediated interference. The collection consists of datasets from 33 wild-type and 1142 RNAi-treated embryos, including five or more datasets for 189 genes. Application of a two-sample t-test identified 8660 reproducible RNAi-induced phenotypes for 421 phenotypic characters. Clustering analysis suggested 24 functional processes essential for early embryogenesis. Our collection is a rich resource for understanding animal development mechanisms.In BriefKyoda et al. used bioimage informatics techniques to create a deep collection of quantitative data on nuclear division dynamics in RNAi-treated C. elegans embryos for 263 essential embryonic genes. Statistical analysis identified 8660 reproducible RNAi phenotypes for 421 phenotypic characters. The collection is a rich resource for understanding animal development.HighlightsBioimage informatics quantified nuclear division dynamics in C. elegans embryosFrom RNAi-silenced embryos we collected 1142 data sets on 263 essential genesStatistical analysis identified 8660 reproducible RNAi phenotypesClustering analysis suggested 24 functional processes in C. elegans embryogenesis

DeepImageJ: A user-friendly plugin to run deep learning models in ImageJ

ABSTRACTDeepImageJ is a user-friendly plugin that enables the generic use in FIJI/ImageJ of pre-trained deep learning (DL) models provided by their developers. The plugin acts as a software layer between TensorFlow and FIJI/ImageJ, runs on a standard CPU-based computer and can be used without any DL expertise. Beyond its direct use, we expect DeepImageJ to contribute to the spread and assessment of DL models in life-sciences applications and bioimage informatics.