News
NCI-Funded Pipeline Applied to Microscopic Images Helps Manage Micronuclei
Are you searching for a better way to investigate micronuclei? There’s a new artificial intelligence (AI)-driven tool, “micronuclAI,” that could help you organize, count, and characterize these tiny structures.
By combining micronuclAI with high-resolution microscopy images, you can isolate and measure even the smallest bits of chromosomal fragments. Those small structures serve as hallmarks of chromosomal instability, a major driver of cancer’s progression and resistance to treatment.
Use the automated pipeline to assess the size, shape, and location of micronuclei in three steps:
- Segmentation
- Micronuclei isolation
- Quantification
Then, once you’ve isolated key portions of the image, you can analyze your data virtually with the help of a convolutional neural network, managed by Heidelberg University. With the resulting micronuclei data, you can make predictions or plots, or you can download the information as a CSV file for use on your local machine.
Dr. Denis Schapiro of Heidelberg University said, “We found micronuclAI performs as well as humans, but in a fraction of the time. For example, it takes about 10 seconds to score 3,000 cells using micronuclAI, compared with about 120 minutes if you do this same task manually.”
He added, “Our pipeline also gave us robust results, even in cases of varying quality, such as images that were out of focus.”
Co-corresponding author, Dr. Benjamin Izar of Columbia University, noted, “In the past, quantifying micronuclei and similar structures to study their underlying biology has been a labor- and resource-intensive task. This new model addresses this issue and gives us a much more efficient way to study these small structures.”
This research is supported in part by NCI's Division of Cancer Biology.