![]() ![]() Exclude the 10X images and non-tiff filesĢ. nd2 files within the dataset that we don’t want to analyze in CellProfiler.Īs an overview, my goals with the input modules were to:ġ. ![]() Each file name also contains this information and contains a separate channel (Phase, GFP, or TxRed). This dataset is structured within folders for the timepoint (hr_6, hr_12, hr_24), the substrate (NP = nonporous, HP05 = 0.5 micron pores, HP3 = 3 micron pores), and the image replicate number (01, 02, 03, etc.). In doing so, the measurements that we eventually make will be linked to an experimental condition. To start, I needed to configure CellProfiler to import the images and extract the relevant experimental metadata for each image. Our collaborators wanted a reproducible pipeline that could be run on dozens of images automatically and tuned by them when needed, so I turned to CellProfiler because it can be adjusted by non-experts and conveniently run on new projects. The HUVECs were grown on three substrates of different porosity and imaged at three different timepoints. Our analysis will focus on the GFP channel only, which captures the connective tissue protein fibronectin labeled with Alexa Fluor® 488 anti-fibronectin antibody. This dataset contains 16-bit images of human umbilical vein endothelial cells (HUVECs) taken at 40X magnification that contain three channels: phase contrast, GFP, and TxRed. The first step in building any pipeline is understanding the underlying data. In a series of several posts, you’ll see each of the tasks that led to a finished pipeline for this project. I find it especially helpful to write down the steps I think I need to do as a way to scope out the project. Sometimes these steps can feel like a winding road, but recognizing where you are in your pipeline building process can help you stay focused and on track. I also performed a background subtraction, which helped to reduce the intensity of everything that wasn’t an object. I masked the debris out, so that nothing in the image was brighter than my objects of interest. If your images don’t look like this initially, you can add processing steps in order to increase the brightness of objects of interest and decrease the brightness of other objects and the background.įor example, with this dataset, I found that many images contained bright debris. In order to accurately segment objects of interest, we need an image where the things that we care about are bright and everything else is dark. Most of these steps are straightforward, with the possible exception of segmentation. A typical project will require importing data, segmenting objects, making measurements, and then exporting the data. The approachĪs you’ll see below, I try to break up my image analysis project into discrete tasks and then work on each of those tasks in turn. in ACS Biomaterials Science & Engineering, where you can learn more about the experiment and results. This dataset was originally published by Chung et al. Their goal for our team was to develop an analysis method to automatically quantify the morphology of fibrils and compare fibril morphology when cells are cultured on substrates of different porosity. Their group uses immunofluorescence microscopy to assay the formation of connective tissue fibers secreted by cells. Tom Gaborski’s NanoBio Materials Laboratory at the Rochester Institute of Technology studies how porous substrates affect the extracellular matrix, which has important consequences for cell migration. In this series of five blog posts, my goal is to share how I approached an image analysis project that I recently undertook as part of the Center for Open Bioimage Analysis. Published by Oxford University Press.One thing I’ve found very helpful in my journey as an image analyst is learning from others by observing their thought process when they tackle a new problem. We implemented an automatic build process that supports nightly updates and regular release cycles for the information: Supplementary data are available at Bioinformatics online. It is available as a packaged application for Mac OS X and Microsoft Windows and can be compiled for Linux. 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).ĬellProfiler Analyst 2.0 is free and open source, available at and from GitHub () under the BSD license. 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. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |