Dr. Eric A. Vitriol. Credit: Michael Holahan, Augusta University
A new “image analysis pipeline” gives researchers a quick new insight into how disease or injury has changed the body, all the way down to the individual cell.
It’s called TDAExplore, which takes the detailed imaging from microscopy, pairs it with a hot field of mathematics called topology, which provides insight into how things are arranged, and the analytical power of artificial intelligence to provide e.g. a new perspective on changes in a cell as a result of ALS and where in the cell they occur, says Dr. Eric Vitriol, cell biologist and neuroscientist at the Medical College of Georgia.
It is an “accessible, powerful option” to use a personal computer to generate quantitative – measurable and consequently objective – information from microscopic images, which could probably also be used for other standard image processing techniques such as X-rays and PET scans, they report in the journal Patterns.
“We think it’s exciting progress to use computers to give us new information about how image sets are different from each other,” says Vitriol. “What are the actual biological changes that are happening, including those that I may not be able to see because they are too small, or because I have a kind of bias around where I need to lead.”
At least in the analytics department, computers beat our brains, the neuroscientist says, not only in their objectivity, but in the amount of data they can assess. Computer vision, which enables computers to extract information from digital images, is a type of machine learning that has been around for decades, so he and his colleague and similar author Dr. Peter Bubenik, a mathematician at the University of Florida and an expert on topological data analysis, decided to collaborate on the details of microscopy with the science of topology and the analytical power of AI. Topology and Bubenik were the key, says Vitriol.
Topology is “perfect” for image analysis because images consist of patterns, of objects arranged in space, he says, and topological data analysis (TDA in TDAExplore) helps the computer also recognize the Earth’s camp, in this case where actin- a protein and essential building blocks in the fibers or filaments that help give cells shape and movement – have moved or changed density. It is an efficient system that instead of taking literally hundreds of pictures to teach the computer to recognize and classify them, it can learn in 20 to 25 pictures.
Part of the magic is that the computer now learns the images in pieces they call patches. Breaking microscopic images into these pieces allows for more accurate classification, less training of the computer in what “normally” looks like, and ultimately extraction of meaningful data, they write.
Microscopy, which allows careful examination of things that are not visible to the human eye, undoubtedly produces beautiful, detailed images and dynamic video that is a cornerstone of many scientists. “You can not have a medical college without sophisticated microscopy facilities,” he says.
But to first understand what is normal and what happens in disease states, Vitriol needs detailed analysis of the images, such as the number of filaments; where the filaments are in the cells – close to the edge, the middle, spread all over it – and whether some cell areas have more.
The patterns that emerge in this case tell him where actin is and how it is organized – an important factor in its function – and where, how and if it has changed with illness or injury.
For example, when he looks at the clustering of actin around the edges of a cell in the central nervous system, the collection tells him that the cell spreads out, moves around, and emits projections that become its leading edge. In this case, the cell, which has essentially been dormant in a bowl, can spread and stretch its legs.
Some of the problems with researchers analyzing the images directly and calculating what they see include that it is time consuming and that even researchers have biases.
As an example, and especially with so much action, their eyes may land on the familiar, in Vitriol’s case, the actin at the leading edge of a cell. When he again looks at the dark frame around the periphery of the cell, which clearly indicates the actin clusters there, it may indicate that it is the most important point of action.
“How do I know that when I decide what’s different, is it the most different thing, or is it just what I wanted to see?” he says. “We want to bring computer objectivity to it, and we want to bring a higher degree of pattern recognition into the analysis of images.”
AI is known to be able to “classify” things, such as recognizing a dog or cat every time, even when the image is blurred, by first learning the many millions of variables associated with each animal, until it knows a dog when it sees one, but it can not report why it is a dog. The approach that requires so many images for training purposes and still does not provide many image statistics does not really work for his purpose, which is why he and his colleagues made a new classification that was limited to topological data analysis.
The bottom line is that the unique linkage used in TDAExplore effectively and objectively tells researchers where and how much the disturbed cell image differs from the training or normal image, which also provides new ideas and research directions, he says.
Back to the cell image showing actin clusters along its perimeter while the “leading edge” was clearly different with perturbations, TDAExplore showed that some of the major changes were actually inside the cell.
“A lot of my job is to try to find patterns in images that are hard to see,” says Vitriol, “because I have to identify those patterns so I can find a way to get numbers out of those images. ” His bottom lines include figuring out how the actin cytoskeleton, for which the filaments form scaffolds and which in turn provides support for neurons, works, and what goes wrong under conditions like ALS.
Some of the machine learning models that require hundreds of images to train and classify images do not describe which part of the image contributed to the classification, the investigators write. Such huge amounts of data to be analyzed and can include 20 million variables require a supercomputer. The new system instead needs relatively few high-resolution images and characterizes the “patches” that led to the chosen classification. In a matter of minutes, the scientist’s standard PC can complete the new image analysis pipeline.
The unique approach used in TDAExplore objectively tells researchers where and how much the disturbed image differs from the training image, information that also provides new ideas and research directions, he says.
The ability to get more and better information from images ultimately means that information generated by basic scientists like Vitriol, who often ultimately changes what is considered facts about a disease and how it is treated, is more accurate. This may include being able to recognize changes, such as those pointed out by the new system inside the cell, which have previously been overlooked.
Currently, researchers are using stains to enable better contrast, and then using software to extract information about what they see in the images, such as how the actin is organized into larger structure, he says.
“We had to find a new way to get relevant data from images, and that’s what this paper is about.”
The published study provides all the parts that other researchers can use TDAExplore.
The Sorting Hat: An AI-powered image classification for cell biologists
Parker Edwards et al., TDAExplore: Quantitative analysis of fluorescence microscopy images through topology-based machine learning, Patterns (2021). DOI: 10.1016 / j.patter.2021.100367
Provided by the Medical College of Georgia at Augusta University
Citation: New method provides quick, objective insight into how cells change by disease (2021, November 23) retrieved November 24, 2021 from https://ift.tt/3nOiCFt disease. html
This document is subject to copyright. Apart from any reasonable trade for the purpose of private investigation or research, no part may be reproduced without written permission. The content is provided for informational purposes only.
Post a Comment