CIF Tutorials | QuPath : Using multiplexed analysis with superpixels

CIF Tutorial : How to use QuPath to create regions with superpixels and classify them like in a multiplex analysis --- [Summary] In this video I'm using QuPath with a fluorescent slide to simulate a multiplexed analysis on regions created via superpixels. The goal is to label these regions according to criteria more complex than just a simple tresholder and get information for each class and sub-class. While the demo only uses two channels, it works with as many channels as you need. Since there is already a good documentation for the basics of QuPath, I will not address them in this video. Some automation and scripting is also shown as an example. PS: Sorry for the loud Outlook chimes and other background noises ... --- [Chapters] 00:00 Intro 01:30 Setting up the project and importing an image into QuPath 02:14 Exploring the image and assessing the situation 03:48 Setting up the classes and creating the training images 05:22 Creating superpixels on one channel 09:46 Adding information to the superpixels 11:40 Creating a classifier for the current channel (red) 17:55 Turning some steps into a script 20:11 Trying to create a classifier for the other channel (green) using a script 22:01 Edit the script to add the missing steps 24:33 Creating the classifier for the other channel (green) 28:19 Creating the superpixels on the original image 29:43 Setting up the classes 31:20 Classifying the superpixels on the original image using a composite classifier 33:56 Getting useful data and comparing regions 37:19 Conclusion [Features used] - Superpixels - Scripting, workflow and automation - Training classifiers - Multiplexed analysis --- [Images] Image from a demo sample (Kidney section) acquired with a Zeiss Axioscan Z1 at the CIF facility. --- [Author] Yannick Krempp, bioimage analyst at the CIF platform, Lausanne, Switzerland https://cif.unil.ch​​ https://www.linkedin.com/in/yannick-k​rempp