Cell segmentation and tracking using CNN-based distance predictions and a graph-based matching strategy

The accurate segmentation and tracking of cells in microscopy image sequences is an important task in biomedical research, e.g., for studying the development of tissues, organs or entire organisms. However, the segmentation of touching cells in images with a low signal-to-noise-ratio is still a challenging problem. In this paper, we present a method for the segmentation of touching cells in microscopy images. By using a novel representation of cell borders, inspired by distance maps, our method is capable to utilize not only touching cells but also close cells in the training process. Furthermore, this representation is notably robust to annotation errors and shows promising results for the segmentation of microscopy images containing in the training data underrepresented or not included cell types. For the prediction of the proposed neighbor distances, an adapted U-Net convolutional neural network (CNN) with two decoder paths is used. In addition, we adapt a graph-based cell tracking algorithm to evaluate our proposed method on the task of cell tracking. The adapted tracking algorithm includes a movement estimation in the cost function to re-link tracks with missing segmentation masks over a short sequence of frames. Our combined tracking by detection method has proven its potential in the IEEE ISBI 2020 Cell Tracking Challenge (http://celltrackingchallenge.net/) where we achieved as team KIT-Sch-GE multiple top three rankings including two top performances using a single segmentation model for the diverse data sets.

Download Full-text

A graph-based cell tracking algorithm with few manually tunable parameters and automated segmentation error correction

10.1101/2021.03.16.435631 ◽

2021 ◽

Author(s):

Katharina Löffler ◽

Tim Scherr ◽

Ralf Mikut

Keyword(s):

Error Correction ◽

Cell Tracking ◽

Automatic Segmentation ◽

Tracking Algorithm ◽

Training Data ◽

Cell Segmentation ◽

Data Sets ◽

False Negatives ◽

Segmentation Error ◽

Tracking Algorithms

AbstractAutomatic cell segmentation and tracking enables to gain quantitative insights into the processes driving cell migration. To investigate new data with minimal manual effort, cell tracking algorithms should be easy to apply and reduce manual curation time by providing automatic correction of segmentation errors. Current cell tracking algorithms, however, are either easy to apply to new data sets but lack automatic segmentation error correction, or have a vast set of parameters that needs either manual tuning or annotated data for parameter tuning. In this work, we propose a tracking algorithm with only few manually tunable parameters and automatic segmentation error correction. Moreover, no training data is needed. We investigate the performance of our approach by simulating erroneous segmentation data, including false negatives, over- and under-segmentation errors, on 2D and 3D cell data sets. We compare our approach against three well-performing tracking algorithms from the Cell Tracking Challenge. Our tracking algorithm can correct false negatives, over- and under-segmentation errors as well as a mixture of the aforementioned segmentation errors. Furthermore, in case of under-segmentation or a combination of segmentation errors our approach outperforms the other tracking approaches.

Download Full-text

A graph-based cell tracking algorithm with few manually tunable parameters and automated segmentation error correction

PLoS ONE ◽

10.1371/journal.pone.0249257 ◽

2021 ◽

Vol 16 (9) ◽

pp. e0249257

Author(s):

Katharina Löffler ◽

Tim Scherr ◽

Ralf Mikut

Keyword(s):

Error Correction ◽

Cell Tracking ◽

Automatic Segmentation ◽

Tracking Algorithm ◽

Training Data ◽

Cell Segmentation ◽

Data Sets ◽

False Negatives ◽

Segmentation Error ◽

Tracking Algorithms

Automatic cell segmentation and tracking enables to gain quantitative insights into the processes driving cell migration. To investigate new data with minimal manual effort, cell tracking algorithms should be easy to apply and reduce manual curation time by providing automatic correction of segmentation errors. Current cell tracking algorithms, however, are either easy to apply to new data sets but lack automatic segmentation error correction, or have a vast set of parameters that needs either manual tuning or annotated data for parameter tuning. In this work, we propose a tracking algorithm with only few manually tunable parameters and automatic segmentation error correction. Moreover, no training data is needed. We compare the performance of our approach to three well-performing tracking algorithms from the Cell Tracking Challenge on data sets with simulated, degraded segmentation—including false negatives, over- and under-segmentation errors. Our tracking algorithm can correct false negatives, over- and under-segmentation errors as well as a mixture of the aforementioned segmentation errors. On data sets with under-segmentation errors or a mixture of segmentation errors our approach performs best. Moreover, without requiring additional manual tuning, our approach ranks several times in the top 3 on the 6th edition of the Cell Tracking Challenge.

Download Full-text

On Improving an Already Competitive Segmentation Algorithm for the Cell Tracking Challenge - Lessons Learned

10.1101/2021.06.26.450019 ◽

2021 ◽

Author(s):

Tim Scherr ◽

Katharina Loeffler ◽

Oliver Neumann ◽

Ralf Mikut

Keyword(s):

Cell Tracking ◽

Data Augmentation ◽

Signal To Noise Ratio ◽

Data Representation ◽

Lessons Learned ◽

Training Data ◽

Fine Tuning ◽

Data Sets ◽

Cell Nuclei ◽

Distance Map

The virtually error-free segmentation and tracking of densely packed cells and cell nuclei is still a challenging task. Especially in low-resolution and low signal-to-noise-ratio microscopy images erroneously merged and missing cells are common segmentation errors making the subsequent cell tracking even more difficult. In 2020, we successfully participated as team KIT-Sch-GE (1) in the 5th edition of the ISBI Cell Tracking Challenge. With our deep learning-based distance map regression segmentation and our graph-based cell tracking, we achieved multiple top 3 rankings on the diverse data sets. In this manuscript, we show how our approach can be further improved by using another optimizer and by fine-tuning training data augmentation parameters, learning rate schedules, and the training data representation. The fine-tuned segmentation in combination with an improved tracking enabled to further improve our performance in the 6th edition of the Cell Tracking Challenge 2021 as team KIT-Sch-GE (2).

Download Full-text

Training a deep learning model for single-cell segmentation without manual annotation

Scientific Reports ◽

10.1038/s41598-021-03299-4 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Nizam Ud Din ◽

Ji Yu

Keyword(s):

Machine Learning ◽

Training Data ◽

Machine Learning Techniques ◽

Cell Segmentation ◽

Learning Techniques ◽

Signal Characteristics ◽

Human Operators ◽

Deep Learning Model ◽

Microscopy Images ◽

Segmentation Models

AbstractAdvances in the artificial neural network have made machine learning techniques increasingly more important in image analysis tasks. Recently, convolutional neural networks (CNN) have been applied to the problem of cell segmentation from microscopy images. However, previous methods used a supervised training paradigm in order to create an accurate segmentation model. This strategy requires a large amount of manually labeled cellular images, in which accurate segmentations at pixel level were produced by human operators. Generating training data is expensive and a major hindrance in the wider adoption of machine learning based methods for cell segmentation. Here we present an alternative strategy that trains CNNs without any human-labeled data. We show that our method is able to produce accurate segmentation models, and is applicable to both fluorescence and bright-field images, and requires little to no prior knowledge of the signal characteristics.

Download Full-text

Unsupervised deep learning method for cell segmentation

10.1101/2021.05.17.444529 ◽

2021 ◽

Author(s):

Nizam Ud Din ◽

Ji Yu

Keyword(s):

Machine Learning ◽

Training Data ◽

Machine Learning Techniques ◽

Cell Segmentation ◽

Learning Techniques ◽

Signal Characteristics ◽

Unsupervised Deep Learning ◽

Human Operators ◽

Microscopy Images ◽

Segmentation Models

Advances in the artificial neural network have made machine learning techniques increasingly more important in image analysis tasks. More recently, convolutional neural networks (CNN) have been applied to the problem of cell segmentation from microscopy images. However, previous methods used a supervised training paradigm in order to create an accurate segmentation model. This strategy requires a large amount of manually labeled cellular images, in which accurate segmentations at pixel level were produced by human operators. Generating training data is expensive and a major hindrance in the wider adoption of machine learning based methods for cell segmentation. Here we present an alternative strategy that uses unsupervised learning to train CNNs without any human-labeled data. We show that our method is able to produce accurate segmentation models. More importantly, the algorithm is applicable to both fluorescence and bright-field images, requiring no prior knowledge of signal characteristics and requires no tuning of parameters.

Download Full-text

Deep learning-based enhancement of epigenomics data with AtacWorks

Nature Communications ◽

10.1038/s41467-021-21765-5 ◽

2021 ◽

Vol 12 (1) ◽

Author(s):

Avantika Lal ◽

Zachary D. Chiang ◽

Nikolai Yakovenko ◽

Fabiana M. Duarte ◽

Johnny Israeli ◽

...

Keyword(s):

Deep Learning ◽

Signal To Noise Ratio ◽

Cell Types ◽

Chromatin Accessibility ◽

Training Data ◽

Regulatory Regions ◽

Hematopoietic Stem ◽

Sequencing Coverage ◽

Lineage Priming ◽

Low Coverage

AbstractATAC-seq is a widely-applied assay used to measure genome-wide chromatin accessibility; however, its ability to detect active regulatory regions can depend on the depth of sequencing coverage and the signal-to-noise ratio. Here we introduce AtacWorks, a deep learning toolkit to denoise sequencing coverage and identify regulatory peaks at base-pair resolution from low cell count, low-coverage, or low-quality ATAC-seq data. Models trained by AtacWorks can detect peaks from cell types not seen in the training data, and are generalizable across diverse sample preparations and experimental platforms. We demonstrate that AtacWorks enhances the sensitivity of single-cell experiments by producing results on par with those of conventional methods using ~10 times as many cells, and further show that this framework can be adapted to enable cross-modality inference of protein-DNA interactions. Finally, we establish that AtacWorks can enable new biological discoveries by identifying active regulatory regions associated with lineage priming in rare subpopulations of hematopoietic stem cells.

Download Full-text

S3norm: simultaneous normalization of sequencing depth and signal-to-noise ratio in epigenomic data

Nucleic Acids Research ◽

10.1093/nar/gkaa105 ◽

2020 ◽

Vol 48 (8) ◽

pp. e43-e43 ◽

Cited By ~ 4

Author(s):

Guanjue Xiang ◽

Cheryl A Keller ◽

Belinda Giardine ◽

Lin An ◽

Qunhua Li ◽

...

Keyword(s):

Gene Expression Regulation ◽

Signal To Noise Ratio ◽

Cell Types ◽

Biological Variation ◽

Sequencing Depth ◽

Data Sets ◽

Data Normalization ◽

Signal To Noise ◽

Experimental Conditions ◽

Multiple Cell

Abstract Quantitative comparison of epigenomic data across multiple cell types or experimental conditions is a promising way to understand the biological functions of epigenetic modifications. However, differences in sequencing depth and signal-to-noise ratios in the data from different experiments can hinder our ability to identify real biological variation from raw epigenomic data. Proper normalization is required prior to data analysis to gain meaningful insights. Most existing methods for data normalization standardize signals by rescaling either background regions or peak regions, assuming that the same scale factor is applicable to both background and peak regions. While such methods adjust for differences in sequencing depths, they do not address differences in the signal-to-noise ratios across different experiments. We developed a new data normalization method, called S3norm, that normalizes the sequencing depths and signal-to-noise ratios across different data sets simultaneously by a monotonic nonlinear transformation. We show empirically that the epigenomic data normalized by our method, compared to existing methods, can better capture real biological variation, such as impact on gene expression regulation.

Download Full-text

A versatile toolbox for semi-automatic cell-by-cell object-based colocalization analysis

Scientific Reports ◽

10.1038/s41598-020-75835-7 ◽

2020 ◽

Vol 10 (1) ◽

Author(s):

Anders Lunde ◽

Joel C. Glover

Keyword(s):

3D Visualization ◽

Image Data ◽

Cell Types ◽

Cellular Heterogeneity ◽

Cell Segmentation ◽

Fluorescence Labeling ◽

Data Sets ◽

Object Based ◽

Imagej Plugin

Abstract Differential fluorescence labeling and multi-fluorescence imaging followed by colocalization analysis is commonly used to investigate cellular heterogeneity in situ. This is particularly important when investigating the biology of tissues with diverse cell types. Object-based colocalization analysis (OBCA) tools can employ automatic approaches, which are sensitive to errors in cell segmentation, or manual approaches, which can be impractical and tedious. Here, we present a novel set of tools for OBCA using a semi-automatic approach, consisting of two ImageJ plugins, a Microsoft Excel macro, and a MATLAB script. One ImageJ plugin enables customizable processing of multichannel 3D images for enhanced visualization of features relevant to OBCA, and another enables semi-automatic colocalization quantification. The Excel macro and the MATLAB script enable data organization and 3D visualization of object data across image series. The tools are well suited for experiments involving complex and large image data sets, and can be used in combination or as individual components, allowing flexible, efficient and accurate OBCA. Here we demonstrate their utility in immunohistochemical analyses of the developing central nervous system, which is characterized by complexity in the number and distribution of cell types, and by high cell packing densities, which can both create challenging situations for OBCA.

Download Full-text