AI improves classic microscopy for soil health testing on farms - Media

American researchers are developing an AI-based microscope system. This could make soil testing faster, cheaper, and more accessible for farmers and land users worldwide. This is reported by UNN with reference to PHYS.org.

Details

Researchers at the University of Texas at San Antonio, USA, have successfully combined low-cost optical microscopy with machine learning to measure the presence and quantity of fungi in soil samples. Their early-stage proof-of-concept technology will be presented at the Goldschmidt Conference in Prague on Wednesday, July 9.

Determining the abundance and diversity of soil fungi can provide valuable information about soil health and fertility, as fungi play an important role in the biogeochemical cycling of nutrients, water retention, and plant growth.

With this knowledge, farmers can optimize crop production and sustainable development by making informed decisions about soil management, including fertilization, irrigation, and tillage.

Optical microscopes are the oldest microscope design and have long been used to detect and identify tiny organisms in soil. Other forms of soil testing use methods such as phospholipid fatty acid testing and DNA analysis to detect organisms or to measure the presence of chemicals such as nitrogen, phosphorus, and potassium.

Leading European business executives ask EU to suspend AI law03.07.25, 13:33 • 6142 views

While these modern methods are powerful, they tend to be expensive or simply emphasize chemical composition, often failing to account for the full biological complexity of soil ecosystems.

Alec Graves from the University of Texas at the College of Sciences San Antonio, USA, is presenting the research at the Goldschmidt Conference this week.

Current forms of biological soil analysis are limited, requiring either expensive laboratory equipment to measure molecular composition or an expert to visually identify organisms using laboratory microscopes. Comprehensive soil testing is not widely available to farmers and landowners who need to understand how agricultural practices affect soil health. By using machine learning algorithms and an optical microscope, we are creating a low-cost soil testing solution that reduces the labor and expertise required, while providing a more complete picture of soil biology.

In the early stage of development, researchers created and tested a machine learning algorithm to detect fungal biomass in soil samples, incorporating it into custom software for labeling microscope images. This was created using a dataset of several thousand images of fungi from soils across South-Central Texas. The software works only with 100x and 400x total microscope magnification, available in many affordable off-the-shelf microscopes, including those in school laboratories.

Our methodology analyzes video of a soil sample, breaks it down into images, and uses a neural network to identify and quantify fungi. Our validation study can already detect fungal hyphae in diluted samples and estimate fungal biomass.

The team is now working on integrating their methodology into a mobile robotic platform for detecting fungi in soil. The system will combine sample collection, microphotography, and analysis in a single device. They aim to have a fully developed, deployable device ready for testing within the next two years.

The research is led by Professor Saugata Datta, Director of the Institute for Sustainable Development and Water Resources Policy at UTSA. Details of the machine learning algorithm are expected to be published in a peer-reviewed journal later this year.

Addition

Google expanded the capabilities of artificial intelligence in Workspace, integrating Gems – individual assistants based on Google Gemini. They are available in Gmail, Docs, Drive, Sheets, and Slides, allowing users to program them for various tasks.