Uncovering the Origin of Animals with AI Microscopy

To answer questions on our biological existence, Prof. Zongjun Yin travels back in time. The Professor of Paleobiology studies the origins of animals: to explore the fossil records within 600-million-year-old sediments. His aim is to reconstruct what the earliest animals looked like: how they reproduced, and how they lived. His quest? To uncover the secrets of the first animal forms that paved the way for the biodiversity we see on Earth today. With the aid of AI microscopy, he peers deeper into these origins and our own place in the tree of life.

Professor Yin began his journey into paleontology: an interdisciplinary field that combines geology and biology. From his doctoral studies at the University of Chinese Academy of Sciences, he delved into the intriguing topic of animal origins. “Emphasizing the importance of researching the origin of animals can never be overstated” he says. “After all, we ourselves are animals, aren't we?” Now, Professor Yin works at NIGPAS, a tight-knit research institute affiliated with the Chinese Academy of Sciences. Comprising of mostly researchers, professors, and engineers, the focus is primarily on fundamental research in paleobiology and stratigraphy. “We all have this innate curiosity to understand who we are and where we are from" he says. “When you trace these questions back, inevitably, you find yourself asking about the very first animals, where they came from, and how they lived.”

Paleontological research in China has a significant standing in the international community. This is due to a relatively large and highly active community of paleontologists. Furthermore, the vast expanse of China's territory has preserved numerous crucial fossil records. For instance, the Chengjiang Biota, dating back 520 million years, has yielded a wealth of soft-bodied animals from the early Cambrian period. This discovery provides a pivotal window into understanding the Cambrian explosion — the rapid diversification of various animal phyla. “Paleontological discoveries in China have helped bridge significant knowledge gaps in the field of evolutionary biology on a global scale” he remarks. Notably, the Chengjiang Biota site is named on the UNESCO World Heritage list, making it the only fossil in Asia to hold this distinction. Another remarkable example is the Jehol Biota, dating back to the dinosaur era. “This site has yielded an astonishing number of feathered dinosaur and bird fossils, offering direct answers to the origins of birds” he says. “Yet these are only a glimpse of China’s many fossil sites.”

Studying the origins of animals helps us to unravel the fundamental question of our biological existence. Theoretically, all animals on Earth, trace back to a common ancestor around 600 million years ago. “My research revolves around exploring the fossil records within these 600-million-year-old sediments” he states. “My aim is to reconstruct what the earliest animals looked like, how they reproduced, and how they lived.” It's a quest to uncover the secrets of the very first animal forms that paved the way for the incredible biodiversity we see on Earth today. Prof. Yin says that understanding these origins not only enriches our scientific knowledge but also provides profound insights into our own place in the tree of life.

“My work involves the challenging task of searching for fossils in the field” he says. “The earliest animals were tiny, often less than 1mm in size and invisible to the naked eye, finding these crucial fossil specimens in the vast wilderness presents a significant challenge.” Most paleontologists will recognize this challenge, which is usually addressed by investing more time into fieldwork.

Yet, some of the more formidable challenges arise in the laboratory. For instance, one of the most significant challenges Dr Yin faces is how to reconstruct the three-dimensional structures of thousands of microfossils rapidly and efficiently. The task is to preserve both the surface morphology and internal structure without damaging the specimens. Additionally, the segmentation and analysis of the volume data obtained from CT scans can be a highly time-consuming and labor-intensive process. Prof. Yin says that analyzing the data from a single specimen can take several weeks, and yet they have thousands of specimens to process. So, finding efficient ways to handle a voluminous dataset quickly and effectively is another major challenge he and his colleagues encounter.

To understand the full picture, Prof. Yin needs to see the internal structures of tiny fossils. Instead of just looking at their surfaces with a regular microscope or SEM, he says he uses high-resolution micro-CT to reconstruct their 3D structures. This way, they can learn about their biology without damaging the specimens. However, his team must scan thousands of specimens to find the internal structures they are looking for. “This used to take weeks or even months for one project, which was very slow” he says, where “many specimens lose their biological information due to decay and diagenetic processes.”

To speed things up, Prof. Yin’s team started using ZEISS DeepRecon, an AI tool which reduces the scanning time without lowering the quality of the data. “For example, a specimen that used to take five hours to scan now takes about one hour or even less” he says “then, we use DeepRecon to process the 3D data.” He asserts that the quality is almost the same as the data he used to get from five hours of scanning. This significantly boosts our efficiency. What used to take a month's work can now be done in a week. When we get enough data in a short time, we can tell a complete and interesting story about the evolution of life. These stories often expand the boundaries of our knowledge. My research is just one example among many in paleontology. AI can help scientists in almost every field by making CT scanning of their specimens faster and more efficient.

Introducing AI into CT scanning has significantly enhanced our efficiency and allowed us to revisit many projects that were previously stalled due to time constraints. With the ability to obtain a large volume of imaging data in a short period, we can now gain crucial insights into the nature of the fossils we study. Take, for example, the Wengan embryo-like fossils. Understanding their position on the tree of life is pivotal to grasping the origins of animals. Imaging data from thousands of specimens is essential for accurately placing them on the tree of life. Without AI, it's hard to imagine having enough data to delve into this issue. Hence, the integration of AI has fundamentally transformed our work approach, accelerating our progress on significant scientific questions.

Moreover, with AI's assistance, researchers can explore a broader array of research avenues. For instance, by scanning numerous insects and plants preserved in amber, scientists can reconstruct the evolution of terrestrial ecosystems. “The possibilities are vast. AI not only accelerates our current research but also opens doors to new horizons,” allowing us to investigate diverse scientific questions that were previously challenging to tackle. My current project is to try to figure out how the first animal on this planet started cell differentiation. If we unravel the answer to this question, we will be one step closer to unraveling the mystery of the origin of animals.