Phenotyping wheat grain using artificial intelligence and X-rays

Imaging innovation harnessed to speed the search for genetic gain in cereals

Innovation
Aa

New X-ray CT scanner set to detect stress tolerance in cereal crop lines.

Aa
FIGURE 1 Rapid computer reconstruction of stress-related damage to wheat ears by an X-ray CT scanner will dramatically reduce screening time for new lines.

FIGURE 1 Rapid computer reconstruction of stress-related damage to wheat ears by an X-ray CT scanner will dramatically reduce screening time for new lines.

Key points

  • An X-ray CT scanner to be housed at the Australian Plant Phenomics Facility will streamline the detection of stress-related damage to wheat grains
  • This rapid phenotyping of hundreds of thousands of heads will greatly accelerate breeding efforts to improve tolerance to frost, heat and drought.

The growing importance of imaging technology to a productive grains industry has been recognised through GRDC investment in an X-ray CT scanner.

The scanner, soon to be installed at the Australian Plant Phenomics Facility (APPF) at the Waite campus of the University of Adelaide, will boost efforts to improve the heat, drought and frost resilience of cereals.

This is an imaging device that uses computer-processed combinations of multiple X-ray measurements - taken from different angles - to produce three-dimensional, cross-sectional images of a scanned object, such as the head of wheat.

Measuring stress damage

The scientific director of the Adelaide node of APPF, Dr Bettina Berger, says that the scanner can help boost stress resilience research by accelerating and refining the analysis of the damage caused to grain.

"Being able to image stress-related damage to grain is what makes it possible to detect those rare lines with unique genetics that are better able to fill grain and maintain yield under stress," Dr Berger says.

Currently, the analysis of harvested wheat spikes from GRDC-supported frost experimental trials is done by hand, requiring months to provide basic measurements on 100,000 heads annually.

The CT scanner - and its associated artificial intelligence (AI) algorithm - dramatically improves throughput, allowing for a more comprehensive analysis of 100,000 heads in possibly as little as six weeks.

The accuracy of measurements is also radically improved, with the scanner producing three-dimensional images at a resolution down to 80 millionths of a metre (80 micrometres).

This amounts to unprecedented phenotyping capability.

The machine is being built in Furth, Germany, at the Fraunhofer Institute in collaboration with Dutch company Phenokey. It is scheduled to be installed in Adelaide in December 2020.

However, work developing the AI algorithm and the methods needed to scan wheat spikes has already been completed and was recently published in scientific literature.

That work was carried out as part of the GRDC investment in the ARC Wheat Hub, where PhD student Jessica Schmidt travelled to the Fraunhofer Institute to carry out X-ray CT scans and then validated the technology by manually phenotyping the scanned heads.

During Ms Schmidt's stay in Germany, heads from 203 highly diverse wheat lines (which were exposed to either drought or combined drought and heat stress) were scanned.

That work, published in the journal Plant Methods, describes a new analytical pipeline that is fundamental to the new system. As well as seed size and seed weight estimates, the method can also measure deformations in grain caused by heat and frost stress.

High-throughput phenotyping of hundreds of thousands of heads will greatly accelerate breeding efforts to improve abiotic stress tolerance. - University of Adelaide scientific director Dr Bettina Berger

Breeding tolerance

"The newly developed pipeline will be scaled up to enable high-throughput phenotyping of hundreds of thousands of heads, greatly accelerating breeding efforts to improve abiotic stress tolerance," Dr Berger says.

GRDC genetic technology officer Dr Hugo Alonso-Cantabrana says there is a strong business case for investment in this imaging capability given that the primary components of yield - grain number and size - are so adversely impacted by environmental stress.

"Current GRDC pre-breeding projects that seek to improve stress tolerance - frost especially - rely heavily on testing grain in thousands of lines annually," Dr Alonso-Cantabrana says.

"Increasing the accuracy and intensity of phenotyping within GRDC-supported projects will accelerate the rate of genetic gain for frost, heat and drought-affected Australian environments."

There is a lot at stake in adding even more resilience to Australian wheat crops. The annual direct and indirect (avoidance strategies) cost of flowering frost and heat stress events to Australian wheat producers is estimated at $1.3 billion.

The X-ray CT scanner is an important part of clawing back those losses.

See also:

GRDC Research Code UOA1806-013AWX

More information: Dr Bettina Berger, 08 8313 0825, bettina.berger@adelaide.edu.au

Aa