One hallmark of many biological processes is cellular phenotypic heterogeneity. Now, a new deep learning method—AINU (artificial intelligence [AI] of the nucleus)—can identify specific nuclear signatures at nanoscale resolution.
Using a small number of images as training data, AINU was able to correctly identify human somatic cells and human induced pluripotent stem cells (hiPSCs). The tool can also differentiate cancer cells from normal cells. In addition, it could detect very early–stage infected cells transduced with DNA herpes simplex virus type 1 (HSV-1) by distinguishing different cell states based on the spatial arrangement of the core histone H3, RNA polymerase II (Pol II), or DNA from super-resolution microscopy images.
The findings are published in the journal Nature Machine Intelligence in an article titled, “A deep learning method that identifies cellular heterogeneity using nanoscale nuclear features.” This work paves the way for improved diagnostic techniques and new monitoring strategies for disease.
AINU is a convolutional neural network that detects and analyzes tiny structures inside cells at the molecular level by scanning high-resolution images obtained using STochastic Optical Reconstruction Microscopy (STORM). STORM reveals structures at nanoscale resolution and can detect rearrangements inside cells as small as 20 nm.
“The resolution of these images is powerful enough for our AI to recognize specific patterns and differences with remarkable accuracy, including changes in how DNA is arranged inside cells, helping spot alterations very soon after they occur,” noted Pia Cosma, PhD, research professor at the . “We think that, one day, this type of information can buy doctors valuable time to monitor disease, personalize treatments, and improve patient outcomes.”
The nanoscale resolution of the images enabled the AI to detect changes in a cell’s nucleus as soon as one hour after it was infected by the herpes simplex virus type-1. The model could detect the presence of the virus by finding slight differences in chromatin structure, which is altered upon viral infection.
“Our method can detect cells that have been infected by a virus very soon after the infection starts. Normally, it takes time for doctors to spot an infection because they rely on visible symptoms or larger changes in the body. But with AINU, we can see tiny changes in the cell’s nucleus right away,” said Ignacio Arganda-Carreras, PhD, a research associate in the computer science and artificial intelligence department at the University of the Basque Country (UPV/EHU).
“Researchers can use this technology to see how viruses affect cells almost immediately after they enter the body, which could help in developing better treatments and vaccines. In hospitals and clinics, AINU could be used to quickly diagnose infections from a simple blood or tissue sample, making the process faster and more accurate,” added Limei Zhong, a researcher at the Guangdong Provincial People’s Hospital (GDPH) in Guangzhou, China.
One hurdle to moving this to the clinic is that STORM imaging typically analyzes only a few cells at a time. For diagnostic purposes, doctors would need to capture many more numbers of cells in a single image to be able to detect or monitor a disease.
“There are many rapid advances in the field of STORM imaging which mean that microscopes may soon be available in smaller or less specialized labs, and eventually, even in the clinic. The limitations of accessibility and throughput are more tractable problems than we previously thought and we hope to carry out preclinical experiments soon,” said Cosma.
The researchers found the technology could identify stem cells with very high precision. AINU can make the process of detecting pluripotent cells quicker and more accurate, helping make stem cell therapies safer and more effective. “Current methods to detect high-quality stem cells rely on animal testing,” said Davide Carnevali, PhD, a p
