Detecting Plant Diseases Earlier Using Hyperspectral Imaging

When it comes to identifying soybean diseases residing in the soil, such as sudden death syndrome (SDS) and brown stem rot, farmers often don’t know of infection until it’s too late to treat or manage them. That’s why Assistant Professor in the Department of Plant Pathology Cory Hirsch is using hyperspectral imaging to detect these diseases that affect farmers’ soybean yields earlier, before the naked eye can detect them.

Cory Hirsch Plant Phenotyping Aurora SporealisTypical digital photography relies on cameras that use red, green, and blue light wavelengths to compose a color image, whereas hyperspectral imaging captures hundreds of wavelengths both in and outside the human visible wavelengths.

“Traditional foliar disease identification for these diseases isn't enough, we need new methods that allow earlier disease detection,” says Hirsch.

The ability to capture greater depth of information allows Hirsch’s team to have a more holistic understanding of when a plant may be infected with or responding to a certain disease. This research is working to identify specific wavelength signatures that indicate if a plant is infected before farmers can visually see it.

One goal of this research is to incorporate and identified wavelength indicators for SDS and brown stem rot into affordable sensor technologies for farmers, allowing them to identify where diseases are present within their fields, which could lead to more precise management decisions.

Plant phenotyping research happening in a soybean field on St. Paul campusAnother goal of this research is to help accelerate and improve plant breeding efforts. A current bottleneck to develop new soybean varieties that respond well in the face of these diseases is that current methods to scan and examine how plants are responding to the disease in the field takes a lot of time, energy and personnel. Leveraging hyperspectral technologies will allow scientists to scan their fields and identify which lines are the best performers more quickly, precisely, and less destructively. This information can be made available to plant breeders so they can decrease the time needed to breed new resistant lines that farmers can use in their fields for more durable yields.

Currently, this technology is being piloted as a way to detect these two specific plant diseases, in the future the Hirsch lab hopes to use hyperspectral imaging and analysis techniques to build more holistic management tools for farmers by helping them have a more complete view of different types of stresses affecting yields, including environmental stresses such as water or nutrient stress.