AI and X-Ray Scans Turn Thousands of Ants Into Stunning 3D Models

A new high-tech scanning system is rapidly turning thousands of ants into stunning 3D models—building a digital library of Earth’s biodiversity.

For more than ten years, Evan Economo’s lab has used micro CT scanners to capture detailed images of insect specimens. These X-ray scans allow scientists to study the shape and structure of insects, an area of entomology called morphology. However, producing these images traditionally requires significant time and expense.

“One limitation is that you can get this rich 3D dataset, but it could take 10 hours to scan one specimen,” explained Economo, chair of the University of Maryland’s Department of Entomology and holder of the James B. Gahan and Margaret H. Gahan Professorship.

To accelerate the process, Economo served as senior author of a study published today (March 5) in Nature Methods. The research team tested a new high-technology workflow designed to dramatically increase scanning speed. Economo worked alongside Thomas van de Kamp of the Karlsruhe Institute of Technology (KIT) in Germany. Their team combined a Synchrotron particle accelerator, X-ray scanning, robotics, and artificial intelligence (AI) to produce interactive digital images representing 800 ant species.

Using this integrated system, the researchers were able to scan specimens far more quickly and convert raw image files into detailed 3D models.

“We’ve estimated that if we were to carry out this project with a lab-based CT scanner, it would take six years of continuous operation,” said Julian Katzke, the study’s first author and a graduate of Economo’s lab at the Okinawa Institute of Science and Technology (OIST) in Japan. “With the setup at KIT, we scanned 2,000 specimens in a single week.”

The effort, called Antscan, could guide future projects aimed at digitizing biological collections. Its data are publicly accessible. Anyone can download the raw files used to create the models, and an integrated viewer allows users to explore the finished 3D ants directly online.

“The value of this study is not only about ants—it’s much broader,” said Economo, who is also an adjunct professor at OIST. “When specimens are digitized, we can build libraries of organisms that can streamline their use from scientific laboratories to classrooms to Hollywood studios.”

Building a Digital Library of Ant Biodiversity

To assemble this large-scale digital collection, the team gathered ethanol-preserved ant specimens from museums, scientific partners, and specialists around the world. After sorting the insects by species and caste, researchers transported them to KIT for high-throughput micro CT scanning. The method works similarly to medical CT imaging but at far higher magnification.

At the facility, a synchrotron particle accelerator generated an intense X-ray beam capable of scanning large numbers of specimens rapidly. A robotic sample changer rotated each specimen and replaced it with the next one every 30 seconds. This process produced stacks of 2D images that could later be combined to create full 3D reconstructions.

The initial scans captured ants in twisted or unnatural positions, which made the models less realistic. To address this, students in UMD Computer Science Associate Professor James Purtilo’s CMSC 435: “Software Engineering” course began developing AI tools to automate “pose estimation.” The system adjusts awkward specimen positions and reconstructs natural postures similar to those ants would display in the wild.

“This collaboration was a great opportunity for us,” Purtilo said. “A capstone is intended to challenge students to integrate skills, function as an effective team, and demonstrate their ability to solve real problems. And this problem was a doozy.”

The resulting Antscan models reveal internal features such as muscles, nervous systems, digestive organs, and stingers with micrometer-level precision. Researchers can animate the models or integrate them into virtual reality environments for scientific studies, educational tools, or visual media.

“To do this manually would have taken years, so without these computational tools it basically would never have been done,” Economo said. “Now, we are making large strides toward creating a living library of interactive models corresponding to Earth’s biodiversity. AI will enable us to explore the diversity of life and share it with the world.”

Antscan Data Drives New Scientific Discoveries

The database is already helping scientists answer new research questions. Economo also served as senior author on a study published in Science Advances on December 19, 2025. In that research, scientists used Antscan data to investigate whether ant colonies benefit more from having many smaller workers or fewer individuals with stronger bodies.

The researchers examined connections between cuticle volume, colony size, and evolutionary diversification in more than 500 ant species. The cuticle forms the protective outer layer of the exoskeleton. Because it requires nitrogen and minerals to produce, thicker armor demands greater resources from the colony for each individual ant.

Their findings revealed a strong negative correlation between cuticle volume and colony size. This suggests that colonies investing in larger numbers of lightly armored workers may be able to grow bigger and diversify more successfully.

Antscan made this type of analysis possible because it allows precise measurement of cuticle volume, something that was previously difficult to quantify. The project also scanned the same species featured in a June 2025 study published in Cell and co-authored by Economo that generated a collection of high-quality ant genomes. Together, these datasets may help scientists explore how physical traits relate to genetic variation.

The extremely detailed scans may also have practical uses in the future. Researchers could train machine learning systems using these images so that algorithms can automatically identify ants in the field during behavioral studies.

Looking ahead, Economo plans to expand the database by scanning additional specimens and continuing collaborations with UMD computer science students to apply AI tools to new biological datasets.

“This work moves us further into the big data era of capturing, analyzing, and sharing organismal shape and form,” Economo said. “The potential for integrating these data with other data types and technologies is immense and very exciting.”

Their paper, “High-throughput phenomics of global ant biodiversity,” was published in the journal Nature Methods on March 5, 2026.

Reference: “High-throughput phenomics of global ant biodiversity” 5 March 2026, Nature Methods.
DOI: 10.1038/s41592-026-03005-0

This article was adapted from text provided by the Okinawa Institute of Science and Technology.

This research was supported by the German Ministry for Research and Education; the Ministry of Science, Research and the Arts Baden-Württemberg; the German Research Foundation (Grant Nos. INST 35/1503-1 FUGG and 502787686); the Okinawa Institute of Science and Technology Graduate University; the Japan Society for the Promotion of Science (Grant Nos. 18K14768, 21K06326 and 22KJ3077); the Australian Research Council (Award No. IC 180100008); HUN-REN Hungarian Research and the National Research, Development, and Innovation Fund (Grant No. K 147781); the Conselho Nacional de Desenvolvimento Científico e Tecnológico (Grant No. 301495/2019-0); the Critical Ecosystem Partnership Fund, a joint initiative of l’Agence Française de Développement, Conservation International, the European Union, the Global Environment Facility, the government of Japan and the World Bank; the U.S. National Science Foundation (Grant Nos. DEB-1932467, DEB 1927161 and IOS-2128304); the Italian Ministry for University and Research; the Environment and Conservation Fund in Hong Kong (Award No. Nb. ECF 137/2020); and Fundação para a Ciência e a Tecnologia. This article does not necessarily reflect the views of these organizations.

Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.