Micrographia@Western is excited to announce our next educational series, the Spring Computational Microscopy Sessions! These afternoon tutorials, running May 6, 7, and 8, will teach you three core skills needed to perform the majority of image analysis tasks – deconvolution, segmentation, and colocalization.
Each session is a stand-alone session, so you can pick and choose which sessions you attend. All three sessions are free, but space is limited, so you must sign up for each session you wish to attend. There are no prerequisites for these sessions, but to get the most out of them you will need to bring a laptop to each session, and you will need to install some free software prior to the sessions. A short description of each day’s module, and signup links, can be found below.
Due to the physics of lenses, all microscopy images are blurred and contain out of focus light. Deconvolution is a collection of image-processing techniques which identify and remove blurred and out-of-focus life from micrographs, improving image resolution, contrast, and quantifiability.
Deconvolution of chlorophyll fluorescence in a cross-section of a stone pine clearly identifies the location of individual chloroplasts
In this session we will discuss the main deconvolution techniques, their advantages and disadvantages, go over approaches to identifying and limiting computational artefacts, discuss the software available in our core facilities, and take you go through a hands-on tutorial of deconvolution using ImageJ, DeconvolutionLab2, and PSFgenerator.
Software Needed:
- ImageJ or FIJI – these are two variants of the same free image analysis software programs. It may be simpler to use ImageJ, but FIJI has greater capabilities.
- DeconvolutionLab2 – an ImageJ/FIJI plugin that provides access to multiple deconvolution techniques.
- PSFgenerator – required for DeconvolutonLab2.
Signup Link: Sorry, registration is closed.
Video Tutorial for Software/Plugin Instillation: Installing FIJI, Deconvolution Lab 2, and PSFgenerator
Day 1: Deconvolution
One of the most common goals in image analysis is to identify and separate different objects in an image based on characteristics such as brightness, shape, size, and texture. Historically, this was a difficult process that often required human annotation. New trainable machine learning algorithms have revolutionized automated segmentation.
Segmentation of macrophages (circled) from collagen autofluorescence in heart tissue stained with anti-CD68.
In this session we will dive into approaches to segmentation using new and emerging machine learning and pre-trained AI tools. These tools greatly simplify this otherwise complex task. We will finish this session with a hands-on tutorial in Ilastik, where we will segment and classify some difficult images.
Software Needed:
- Ilastik – a free and open source machine learning tool for segmentation and other pixel classification tasks.
- FIJI – a free image processing software package
- Optional – if time permits we will go through an example of μSAM. This is a more capable segmentation system, but in also harder to install and run. If you are comfortable working in a Python Conda environment, than we encourage you to install μSAM and follow along with this demo:
- Python + Conda
- Napari Image Processing Environment
- μSAM (Napari plugin)
Signup Link: Sorry, registration is closed.
Video Tutorial for Software/Plugin Instillation: Installing ilastik and Linking it to FIJI
Day 2: Pixel Classification and Segmentation
Micrographs can be used to detect the co-expression of proteins in cells, the accumulation of markers on the same subcellular structures, and even test for the presence of inter-protein or inter-object interactions. The techniques for quantifying these forms of interaction are collectively known as colocalization, and when used properly, can be a powerful analytical tool.
Pearson’s colocalization analysis (right) of the interaction between microtubules (magenta) and actin (yellow) in epithelial cells (left).
In this session we will discuss several colocalization metrics, their use, and their interpretation. You will learn how to employ these methods in a statistically robust fashion. Each method will be applied in a series of hands-on exercises using the JACoP colocalization tool in ImageJ.
Software Needed:
- ImageJ or FIJI – these are two variants of the same free image analysis software programs. It may be simpler to use ImageJ, but FIJI has greater capabilities.
- Just Another Colocalization Plugin (JACoP) – this is an excellent colocalization plugin for ImageJ/FIJI and it has an accompanying paper to help you understand each analysis.
Signup Link: Sorry, registration is closed.
Video Tutorial for Software/Plugin Instillation: Installing FIJI and Just Another Colocalization Plugin