An AI-based image classifier for cell biologists

Applying deep learning artificial intelligence (AI) for image classification requires a high level of machine learning expertise. Today, Japanese researchers have created an AI image classifier that can be customized and easily used by non-experts. Credit: Kiyotaka Nagaki from Okayama University

Cell division is an important process underlying biological growth and repair. Cell biologists follow this process by looking at chromosomes, structures made up of DNA that contain the genetic material of an organism. Advances in microscopy as well as automation have allowed researchers to take better images of chromosomes in a short time. However, much of their analysis is still done manually, which is often a tedious task. This is especially true for plants, which exhibit a wide variety of sizes and chromosome numbers.

However, in a recent study published in Chromosome research, Japanese researchers took a different approach. Led by Associate Professor Kiyotaka Nagaki of Okayama University, Japan, they used deep learning artificial intelligence (AI) to classify chromosome images of several plant species. While this in and of itself is nothing new, what is interesting is that the team has shown that it is possible, even for non-experts, to use AI easily.

How was this possible? Dr Nagaki says: “Classifying images using AI generally requires a high level of computer literacy. What we did was create AI models on a McIntosh computer with the CreateML application tailored to our own sample images. a custom image classifier for any variety of images that suit its purpose. “

The team used chromosomal images to train deep learning models to detect images or parts of images where cells undergo “mitosis,” a process in which a single cell divides into two daughter cells. identical. They estimated its detection accuracy with test images based on the number of cells correctly classified by the model.

Next, the team tested the models with images containing mitotic cells from plant species not used during training. To their delight, the models correctly distinguished the mitotic cells in these images. Additionally, the technique also worked well for cells in tissue sections and a different cell division process.

These results indicate that the deep learning pipeline developed by the team can be used easily and reliably by non-data scientists in different disciplines, greatly simplifying and speeding up the task of image analysis.

In addition, the scope of this reported method can be extended to more complex analyzes such as the identification of chromosomal aberrations and the development of new advisory systems for the detection and classification of objects. “There are more trivial classifications in our lives than one might imagine. Automating such classifications by handing them over to an AI can not only eliminate fluctuations caused by individual differences, but also save many hours. Streamlining such trivial classifications allows for in-depth image-based studies more reproducible and reliable, ”says Dr Nagaki.

Indeed, “a deep learning sorter that anyone can use,” as he puts it, could be our key to understanding the different nuances of biological species.

Convolutional neural network used to identify dog ​​breeds from photographs

More information:
Kiyotaka Nagaki et al, Effectiveness of Create ML in microscopic image classifications: a simple and inexpensive deep learning pipeline for non-data scientists, Chromosome research (2021). DOI: 10.1007 / s10577-021-09676-z

Provided by Okayama University

Quote: The Sorting Hat: An AI-Based Image Classifier for Cell Biologists (2021, November 16) Retrieved November 16, 2021 from -image-cell-biologists .html

This document is subject to copyright. Other than fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for information only.

Source link

Comments are closed.