DeepSynth: Three-dimensional nuclear segmentation of biological images using neural networks trained with synthetic data

Abstract

The scale of biological microscopy has increased dramatically over the past ten years, with the development of new modalities supporting collection of high-resolution fluorescence image volumes spanning hundreds of microns if not millimeters. The size and complexity of these volumes is such that quantitative analysis requires automated methods of image processing to identify and characterize individual cells. For many workflows, this process starts with segmentation of nuclei that, due to their ubiquity, ease-of-labeling and relatively simple structure, make them appealing targets for automated detection of individual cells. However, in the context of large, three-dimensional image volumes, nuclei present many challenges to automated segmentation, such that conventional approaches are seldom effective and/or robust. Techniques based upon deep-learning have shown great promise, but enthusiasm for applying these techniques is tempered by the need to generate training data, an arduous task, particularly in three dimensions. Here we present results of a new technique of nuclear segmentation using neural networks trained on synthetic data. Comparisons with results obtained using commonly-used image processing packages demonstrate that DeepSynth provides the superior results associated with deep-learning techniques without the need for manual annotation.

Pomegranate: 2D segmentation and 3D reconstruction for fission yeast and other radially symmetric cells

“…nuclei or cells). Segmentation can be achieved by traditional image processing algorithms such as adaptive thresholding and binary morphometric operations, or by machine learning or deep learning techniques [4][5][6][7][8][9][10][11][12][13][14] . However, any strategy that uses a fluorescent marker for live-cell segmentation (1) increases phototoxicity, (2) limits the number of fluorescent channels available for proteins or structures of interest, and (3) can cause variability in segmentation when uneven illumination, or uneven staining or expression, leads to variability in fluorescence signal.…”

A bird’s-eye view of deep learning in bioimage analysis

“…An alternative approach is transfer learning between domains [255] , [278] , [286] which holds great promise for biomedical imaging as well [287] , [288] , [289] , [290] . Another strategy popular in bioimage analysis is to use high-fidelity simulated data as a surrogate for real data [256] , [291] , [292] , which allows supervised learning with any number of images without requiring manual annotation [293] , [294] , [295] .…”

In situclassification of cell types in human kidney tissue using 3D nuclear staining

“…12,13,16,31,32 Many efforts are also focused on cell segmentation, automated detection and counting. 33 Digital pathology relies on histology staining and has experienced exciting development in cell segmentation and classification. 8,[34][35][36] Several approaches have been described to classify cells based only on nuclear staining in digital pathology images, such as standard machine learning classifiers, numerical feature engineering, neural networks and transport-based morphometry.…”