Phenotype pictures of wheat heads infected with Fusarium graminearum

Fusarium head blight (FHB) of wheat (Triticum aestivum), caused by the fungal pathogen Fusarium graminearum (Fg), reduces grain yield and quality due to the production of the mycotoxin deoxynivalenol. Manual rating for incidence (percent of infected wheat heads/spikes) and severity (percent of spikelets infected) to estimate FHB resistance, is time-consuming and subject to human error. This study uses a deep learning model, combined with a spectral index, to provide rapid phenotyping of FHB severity. An object detection model was used to localize wheat heads within boundary boxes. Corresponding boxes were used to prompt Meta’s Segment Anything Model to segment wheat heads. Using 2576 images of wheat heads point inoculated with Fg in a controlled environment, a spectral index was developed using the red and green bands to differentiate healthy from infected tissue and estimate disease severity. Stratified random sampling was applied to pixels within the segmentation mask, and the model classified pixels as healthy or infected with an accuracy of 87.8%. Linear regression determined the relationship between the index and visual severity scores. The severity estimated by the index was able to predict visual scores (R2=0.83, p=<2e-16). This workflow was also applied to plot size images of infected wheat heads from an outside dataset with varying cultivars and light conditions, to assess model transferability. It correctly classified pixels as healthy or infected with a prediction accuracy of 85.8%. These methods may provide rapid estimation of FHB severity to improve selection efficiency for resistance or estimate disease pressure for effective management.

Methods
Single wheat heads from six different winter wheat varieties were point inoculated with Fusarium graminearum and kept in a controlled environment optimal for fungal growth. Images were taken of each wheat head 7, 10, and 14 days post-inoculation. Each image was taken with a standard iPhone 14 pro camera with automatic settings under consistent lighting conditions, with the head placed upon a black background. Each Image was taken from approximately 30 cm. In addition, visual disease severity notes were taken for each head by counting the number of infected spikelets on each head at each day an image was taken. Two approaches were then used to estimate severity within the heads, a deep learning only approach and a deep learning combined with a spectral index approach. 

A gain and exposure test was then conducted to determine the effect of camera settings on the workflow. Images were taken of the same infected wheat plants under varying outdoor light conditions, some with automatic gain and exposure camera settings and others with set settings. Annotations were then created for the wheat heads within the images, and index values calculated to determine thresholds for these images.