Geropathology Imaging - Part of the The Jackson Laboratory Shock Center of Excellence in the Basic Biology of Aging
How to use Aging Classifiers
We provide directions on how to make your own at the end
Concepts
Understanding what machine learning is going is sometimes difficult to understand. you won’t need to code or to know the details of machine learning to use our tools but some good background can be found here.
Link to basic concepts video
Link to Flow chart of work
Overall Workflow
The chart here overall shows you how to get from Images to Aging scores. We downsample the slides using Qupath then we use our classifers with Julia to extract Zarrs from the tiles then we use these files in the classifer to put and age score we then move to a Jupyter notebook to get data and do some statistics
Coding Basics
We want to support all levels of users. Coding was new to us when we started this process and understand that it can be intimidating. We hope that these recourses will provide confidence to new users.
Link to coding basicis
Links to Required Software
Required Software
There are 4 different pieces of software we utalize to go from Images to ageing scores. The good news is that all of these are free! Follow the link to get to thier websites and installation directions
How to use our classifier:
While there are several ways to get to the downsampled tiles we extract from Qupath we provide instructions on how to get to these tiles assuming your storing your images in Omero.
We use Qupath for this as it is easy to use and interactive. Click on the link for directions on how to get Images from Omero to Qupath.
Once you have your Images in Qupath you will need to annotate them. Annotate slides in Qupath After they are annotated we can use Qupath to export downsampled tilesexporting-tiles-from-qupath
The tiles then need to be stiched together, This will be the first time we will be using jupyter notebook to use some code. Directions can be found here stitching-tiles
After the images are stitched, we will then run them through a classifier. We will use Julia to Run a classifier.
Once the classifer has been run we need to compile the raw age scores. To do this we switch back to jupyter notebooks to convert Raw Age Scores to Compiled Age Scores, With the tissue scores in hand, we have the option of painting these scores as a way of visualizing the aged classifier results.. The results can also be complied into a csv file for statistsics LINK We provide some statistics and graphs we think might be helpful for many users in R. LINK