At Cambridge, the search for plants that resist root-invading nematodes has acquired a much faster pair of eyes. Researchers report that 3D-printed robots and AI image analysis have lifted screening capacity from about 90 plants a day to around 1,000 plants an hour.[1][2]
The machines tackle a bottleneck in crop research: observing microscopic parasites as they infect roots. Repeating those observations over several days also lets the team follow an infection’s course, rather than reducing each plant to a single final count.[1][2]
More observations can help reveal which plant genotypes impede a parasite and when that defence acts. The immediate advance is research capacity; connecting the measurements to durable resistance in food crops remains the biological task.
A biological battle under the surface
What the screening system adds
| Capability | Research use |
|---|---|
| Around 1,000 plants per hour | Reported laboratory screening capacity |
| Repeated imaging over days | Follow how an infection develops |
| Automated image analysis | Compare parasite responses across plant genotypes |
Not all nematodes are crop pests. Many are free-living organisms with important ecological roles. Plant-parasitic species, however, can enter root tissues, establish feeding sites and affect the plant's capacity to take up water and nutrients. Damage may be difficult to identify from surface appearance alone until yield or growth has already suffered.[1][2]
A resistance-screening experiment needs to distinguish among plants that exclude a parasite, plants that tolerate its presence, and plants in which the parasite feeds or reproduces successfully. Those are related but different biological outcomes. Counting the number and size of nematodes, correlating their behaviour with the host genotype, and observing a sequence of changes may reveal more than a binary label of healthy or unhealthy.
Traditional assessment can involve painstaking manual preparation, microscopy and image annotation. It is difficult to scale those routines across hundreds of plant genotypes and tens of thousands of parasites. Each observation is also a snapshot. If two plants carry the same number of worms at one moment but differ sharply in how those worms develop, a single count could hide an important resistance mechanism.[1][2]
A four-dimensional view of a root infection
The group describes the system as enabling four-dimensional imaging: observing spatial structure as it changes through time. Printed hardware supports repeatable image collection, while computer vision helps analyse the plant–parasite interactions across many samples.
The researchers are currently using model plants to study these interactions. Their next step is to apply the system to food crops such as potatoes, bringing the faster observations closer to the breeding problems the project aims to address.[1]
Automated analysis carries its own potential failure modes. A model can confuse a worm with a root structure, fail on poorly focused images or work better on some genotypes than others. Repeated tracking must also recognise the same parasite across different images as it shifts position or changes appearance. A credible breeding programme needs independent validation of those measurements, particularly for candidate plants selected as resistant.[1][2]
What faster observations can reveal
The attraction is the ability to ask a more detailed question of each candidate plant: not simply whether parasites are present, but how the interaction develops. A resistance mechanism that would be missed by one snapshot may become visible in a sequence. That is the route from faster screening to more informative crop genetics.[1][2]
Sources & further reading
- AI-powered nematode imaging for global food security University of Cambridge, Department of Plant Sciences · 12 Oct 2026 · Accessed 26 Oct 2026