AI-Boosted Detection of Biogenicity in Sediments

Published Date : 24/10/2025 

Scientists have developed an AI classifier to detect biogenicity in sedimentary structures, which could have significant implications for identifying signs of past life on Mars. 

The search for traces of life can be based on the detection of specific signatures produced by microorganisms on sedimentary rocks. These signatures, known as microbially induced sedimentary structures (MISSs), develop under specific physicochemical conditions that are likely to have existed on Mars during the Noachian period.

We designed an experiment under controlled laboratory conditions to explore the wide range of variability in biogeomorphological responses of clay-sand substrates to the development of biological mats, including microbial mats, of different strains and biomasses, and an abiotic control. This experiment aimed to mimic the conditions that could have been present on early Mars, providing insights into the potential for microbial life on the red planet.

A 3D picture dataset based on the experiment was built using multi-image photogrammetry. Visual observations were combined with multivariate statistics on computed topographical variables to interpret the diversity in the resulting biotic and abiotic mud cracks. This approach allowed for a detailed analysis of the structural differences between biotic and abiotic samples, which is crucial for understanding the potential for microbial life on other planets.

Finally, an artificial intelligence (AI) classifier based on convolutional neural networks was trained with the data. The resulting model predicted accurately not only the biotic-abiotic differences but also the differences between strains and biomasses of biotic treatments. Its results outperformed the blind human classification, even using only grayscale pictures. This suggests that AI can be a powerful tool in the detection of biogenicity in sedimentary structures, potentially enhancing our ability to identify signs of past life on Mars.

Class Activation Maps showed that AI followed several decision paths, not always like those of the human expert. This indicates that AI can provide new insights and perspectives that might be missed by human analysis. The next steps are proposed for the application of these models to ex situ biogeomorphological structures (fossil and modern MISS) on Earth’s surface, to ultimately transpose them to a martian context.

The implications of this research are significant. By using AI to enhance the detection of biogenicity in sedimentary structures, scientists can improve their ability to identify potential signs of past life on Mars. This could lead to more targeted and effective missions to explore the red planet, potentially uncovering evidence of ancient microbial life. 

Frequently Asked Questions (FAQS): 

Q: What are microbially induced sedimentary structures (MISSs)?

A: Microbially induced sedimentary structures (MISSs) are specific signatures produced by microorganisms on sedimentary rocks. These structures form under specific physicochemical conditions and can be used to identify the presence of past microbial life.

Q: How was the experiment designed?

A: The experiment was designed to explore the biogeomorphological responses of clay-sand substrates to the development of biological mats, including microbial mats, of different strains and biomasses, under controlled laboratory conditions. An abiotic control was also included for comparison.

Q: What techniques were used to analyze the data?

A: A 3D picture dataset was created using multi-image photogrammetry. Visual observations were combined with multivariate statistics on computed topographical variables to interpret the diversity in the resulting biotic and abiotic mud cracks.

Q: What is the role of AI in this research?

A: An AI classifier based on convolutional neural networks was trained to predict the biotic-abiotic differences and the differences between strains and biomasses of biotic treatments. The AI outperformed blind human classification, even using only grayscale pictures.

Q: What are the next steps for this research?

A: The next steps include applying these models to ex situ biogeomorphological structures (fossil and modern MISS) on Earth’s surface, with the ultimate goal of transposing them to a martian context to enhance the detection of biogenicity in sedimentary structures. 

More Related Topics :