Quantitative analysis of cell organelles with artificial intelligence

Bessie II’s high-sensitivity X-rays can be used to create microscopic images with a spatial resolution down to a few tens of nanometers. Entire cell blocks can be examined without the need for complex sample preparation as in electron microscopy. Under the X-ray microscope, tiny cell organelles with microscopic structures and boundary membranes appear clear and detailed, even in three dimensions.

This makes cryo-X-ray tomography ideal for, for example, studying changes in cell structures caused by external stimuli. However, until now, evaluation of 3D tomograms often required manual and labor-intensive data analysis. To tackle this problem, teams led by computer scientist Prof. Dr. Frank Noh and cell biologist Prof. Dr. Helge Evers (both from Freie Universität Berlin) have now collaborated with the Department of X-ray Microscopy at the HZB.

A computer science team has developed a novel, self-learning algorithm. This AI-based analysis method is based on automatic detection of subatomic structures and accelerates quantitative analysis of 3D X-ray data sets. 3D images of the interior of biological samples were acquired at the U41 beamline at BESSY II.

„In this study, using mammalian cells from cell cultures called filopodia, we have now shown how well AI-based analysis of cell volumes works,” says Dr. Stephan Werner, X-ray microscope expert at HZB. . Mammalian cells have a complex structure with various organelles, each of which must fulfill different cellular functions. Filopodia are protrusions of the cell membrane and specifically aid in cell migration.

„For cryo-X-ray microscopy, the cell samples are first shock-frozen, so ice crystals do not form inside the cell. This leaves the cells in an almost natural state and allows us to study the influence of external factors on the structure of the cell,” explains Werner.

„Our work has already sparked considerable interest among experts,” says first author Michael Dier from the Freie Universität Berlin. The neural network correctly recognizes 70% of the existing cell features within a very short time, thus enabling a very fast evaluation of the data set. „In the future, we can use this new analysis method to investigate how cells react to environmental influences such as nanoparticles, viruses or carcinogens much faster and more reliably than before,” Diehr says.

The work has been published in the Journal Proceedings of the National Academy of Sciences.