Keras R-CNN: library for cell detection in biological images using deep neural networks.

BACKGROUND: A common yet still manual task in basic biology research, high-throughput drug screening and digital pathology is identifying the number, location, and type of individual cells in images. Object detection methods can be useful for identifying individual cells as well as their phenotype in one step. State-of-the-art deep learning for object detection is poised to improve the accuracy and efficiency of biological image analysis. RESULTS: We created Keras R-CNN to bring leading computational research to the everyday practice of bioimage analysts. Keras R-CNN implements deep learning object detection techniques using Keras and Tensorflow ( https://github.com/broadinstitute/keras-rcnn ). We demonstrate the command line tool's simplified Application Programming Interface on two important biological problems, nucleus detection and malaria stage classification, and show its potential for identifying and classifying a large number of cells. For malaria stage classification, we compare results with expert human annotators and find comparable performance. CONCLUSIONS: Keras R-CNN is a Python package that performs automated cell identification for both brightfield and fluorescence images and can process large image sets. Both the package and image datasets are freely available on GitHub and the Broad Bioimage Benchmark Collection.

Data-analysis strategies for image-based cell profiling.

Image-based cell profiling is a high-throughput strategy for the quantification of phenotypic differences among a variety of cell populations. It paves the way to studying biological systems on a large scale by using chemical and genetic perturbations. The general workflow for this technology involves image acquisition with high-throughput microscopy systems and subsequent image processing and analysis. Here, we introduce the steps required to create high-quality image-based (i.e., morphological) profiles from a collection of microscopy images. We recommend techniques that have proven useful in each stage of the data analysis process, on the basis of the experience of 20 laboratories worldwide that are refining their image-based cell-profiling methodologies in pursuit of biological discovery. The recommended techniques cover alternatives that may suit various biological goals, experimental designs, and laboratories' preferences.

Assessing microscope image focus quality with deep learning.

BACKGROUND: Large image datasets acquired on automated microscopes typically have some fraction of low quality, out-of-focus images, despite the use of hardware autofocus systems. Identification of these images using automated image analysis with high accuracy is important for obtaining a clean, unbiased image dataset. Complicating this task is the fact that image focus quality is only well-defined in foreground regions of images, and as a result, most previous approaches only enable a computation of the relative difference in quality between two or more images, rather than an absolute measure of quality. RESULTS: We present a deep neural network model capable of predicting an absolute measure of image focus on a single image in isolation, without any user-specified parameters. The model operates at the image-patch level, and also outputs a measure of prediction certainty, enabling interpretable predictions. The model was trained on only 384 in-focus Hoechst (nuclei) stain images of U2OS cells, which were synthetically defocused to one of 11 absolute defocus levels during training. The trained model can generalize on previously unseen real Hoechst stain images, identifying the absolute image focus to within one defocus level (approximately 3 pixel blur diameter difference) with 95% accuracy. On a simpler binary in/out-of-focus classification task, the trained model outperforms previous approaches on both Hoechst and Phalloidin (actin) stain images (F-scores of 0.89 and 0.86, respectively over 0.84 and 0.83), despite only having been presented Hoechst stain images during training. Lastly, we observe qualitatively that the model generalizes to two additional stains, Hoechst and Tubulin, of an unseen cell type (Human MCF-7) acquired on a different instrument. CONCLUSIONS: Our deep neural network enables classification of out-of-focus microscope images with both higher accuracy and greater precision than previous approaches via interpretable patch-level focus and certainty predictions. The use of synthetically defocused images precludes the need for a manually annotated training dataset. The model also generalizes to different image and cell types. The framework for model training and image prediction is available as a free software library and the pre-trained model is available for immediate use in Fiji (ImageJ) and CellProfiler.

An open-source solution for advanced imaging flow cytometry data analysis using machine learning.

Imaging flow cytometry (IFC) enables the high throughput collection of morphological and spatial information from hundreds of thousands of single cells. This high content, information rich image data can in theory resolve important biological differences among complex, often heterogeneous biological samples. However, data analysis is often performed in a highly manual and subjective manner using very limited image analysis techniques in combination with conventional flow cytometry gating strategies. This approach is not scalable to the hundreds of available image-based features per cell and thus makes use of only a fraction of the spatial and morphometric information. As a result, the quality, reproducibility and rigour of results are limited by the skill, experience and ingenuity of the data analyst. Here, we describe a pipeline using open-source software that leverages the rich information in digital imagery using machine learning algorithms. Compensated and corrected raw image files (.rif) data files from an imaging flow cytometer (the proprietary .cif file format) are imported into the open-source software CellProfiler, where an image processing pipeline identifies cells and subcellular compartments allowing hundreds of morphological features to be measured. This high-dimensional data can then be analysed using cutting-edge machine learning and clustering approaches using "user-friendly" platforms such as CellProfiler Analyst. Researchers can train an automated cell classifier to recognize different cell types, cell cycle phases, drug treatment/control conditions, etc., using supervised machine learning. This workflow should enable the scientific community to leverage the full analytical power of IFC-derived data sets. It will help to reveal otherwise unappreciated populations of cells based on features that may be hidden to the human eye that include subtle measured differences in label free detection channels such as bright-field and dark-field imagery.

CellProfiler Analyst: interactive data exploration, analysis and classification of large biological image sets.

CellProfiler Analyst allows the exploration and visualization of image-based data, together with the classification of complex biological phenotypes, via an interactive user interface designed for biologists and data scientists. CellProfiler Analyst 2.0, completely rewritten in Python, builds on these features and adds enhanced supervised machine learning capabilities (Classifier), as well as visualization tools to overview an experiment (Plate Viewer and Image Gallery). AVAILABILITY AND IMPLEMENTATION: CellProfiler Analyst 2.0 is free and open source, available at http://www.cellprofiler.org and from GitHub (https://github.com/CellProfiler/CellProfiler-Analyst) under the BSD license. It is available as a packaged application for Mac OS X and Microsoft Windows and can be compiled for Linux. We implemented an automatic build process that supports nightly updates and regular release cycles for the software. CONTACT: anne@broadinstitute.orgSupplementary information: Supplementary data are available at Bioinformatics online.

Piximi - An Images to Discovery web tool for bioimages and beyond.

Deep learning has greatly accelerated research in biological image analysis yet it often requires programming skills and specialized tool installation. Here we present Piximi, a modern, no-programming image analysis tool leveraging deep learning. Implemented as a web application at Piximi.app, Piximi requires no installation and can be accessed by any modern web browser. Its client-only architecture preserves the security of researcher data by running all computation locally. Piximi offers four core modules: a deep learning classifier, an image annotator, measurement modules, and pre-trained deep learning segmentation modules. Piximi is interoperable with existing tools and workflows by supporting import and export of common data and model formats. The intuitive researcher interface and easy access to Piximi allows biological researchers to obtain insights into images within just a few minutes. Piximi aims to bring deep learning-powered image analysis to a broader community by eliminating barriers to entry.

Applying Faster R-CNN for Object Detection on Malaria Images.

Deep learning based models have had great success in object detection, but the state of the art models have not yet been widely applied to biological image data. We apply for the first time an object detection model previously used on natural images to identify cells and recognize their stages in brightfield microscopy images of malaria-infected blood. Many micro-organisms like malaria parasites are still studied by expert manual inspection and hand counting. This type of object detection task is challenging due to factors like variations in cell shape, density, and color, and uncertainty of some cell classes. In addition, annotated data useful for training is scarce, and the class distribution is inherently highly imbalanced due to the dominance of uninfected red blood cells. We use Faster Region-based Convolutional Neural Network (Faster R-CNN), one of the top performing object detection models in recent years, pre-trained on ImageNet but fine tuned with our data, and compare it to a baseline, which is based on a traditional approach consisting of cell segmentation, extraction of several single-cell features, and classification using random forests. To conduct our initial study, we collect and label a dataset of 1300 fields of view consisting of around 100,000 individual cells. We demonstrate that Faster R-CNN outperforms our baseline and put the results in context of human performance.

Exhaled carbon monoxide level as an indicator of cigarette consumption in a workplace cessation program in Taiwan.

BACKGROUND: Smoking cessation programs are critical to the safety and health of workers. Exhaled carbon monoxide (CO) is an effective indicator of smoking in clinics and hospitals. Its application in the community and workplace, however, remains limited. This study assessed whether exhaled CO concentration can be used as an objective indicator of the amount of daily cigarette consumption among smokers in the workplace in Taiwan. METHODS: A total of 150 workers from a chemical manufacturer in Taiwan were included; there were 27 nonsmokers and 123 current smokers. The number of cigarettes smoked daily by each subject was reported, and exhaled CO concentration was measured in each subject using the Micro CO meter (Micro Medical Ltd, Chatham, Kent, UK). RESULTS: Exhaled CO levels were associated with the number of cigarettes consumed daily, with a correlation coefficient of 0.73 (p < 0.01) and an adjusted R-square (simple linear regression model) of 0.44. The mean exhaled CO level of nonsmokers was 4.2 ppm (95% confidence interval, 3.3-5.1). A reading of > 6 ppm had a sensitivity of 84% and specificity of 85% in detecting workplace smoking. CONCLUSION: Exhaled CO level can be used as an objective, noninvasive indicator to determine the smoking status of an individual in the workplace.

Eigenspace Update for Molecular Geometry Optimization in Nonredundant Internal Coordinate.

An eigenspace update method is introduced in this article for molecular geometry optimization. This approach is used to obtain the nonredundant internal coordinate space and diagonalize the Hessian matrix. A select set of large molecules is tested and compared with the conventional method of direct diagonalization in redundant space. While all methods considered herein take on similar optimization pathways for most molecules tested, the eigenspace update algorithm becomes much more computationally efficient with increasing size of the molecular system. A factor of 3 speed-up in overall computational cost is observed in geometry optimization of the 25-alanine chain molecule. The contributing factors to the computational savings are the reduction to the much smaller nonredundant coordinate space and the O(N(2)) scaling of the algorithm.

Effects of a home program on strength, walking speed, and function after total hip replacement.

OBJECTIVE: To assess the efficacy of a home exercise program in increasing hip muscle strength, walking speed, and function in patients more than 1.5 years after total hip replacement (THR). DESIGN: Randomized controlled trial. SETTING: Kinesiology laboratory. PARTICIPANTS: Fifty-three patients with unilateral THR were randomly assigned to the training (n=26) and control (n=27) groups. Patients in the training group were further divided into exercise-high (n=13) and exercise-low (n=13) compliance groups according to their practice ratio (high, > or =50%). INTERVENTION: The training group underwent a 12-week home program that included hip flexion range of motion exercises for both hip joints; strengthening exercises for bilateral hip flexors, extensors, and abductors; and a 30-minute walk every day. The control group did not receive any training. MAIN OUTCOME MEASURES: Strength of bilateral hip muscles, free and fast walking speeds while walking over 3 different terrains, and functional performance were assessed by using a dynamometer, videotape analysis, and the functional activity part of the Harris Hip Score, respectively, before and after the 12-week period. RESULTS: Subjects in the exercise-high compliance group showed significantly (P <.05) greater improvement in muscle strength for the operated hip, fast walking speed, and functional score than those in the exercise-low compliance and control groups. CONCLUSIONS: The designed home program was effective in improving hip muscle strength, walking speed, and function in patients after THR who practiced the program at least 3 times a week, but adherence to this home program may be a problem.