|DIVILOV, KONSTANTIN - Cornell University|
|WIESNER-HANKS, TYR - Cornell University|
|BARBA, PAOLA - Cornell University|
|REISCH, BRUCE - Cornell University|
Submitted to: Phytopathology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/22/2017
Publication Date: 9/28/2017
Citation: Divilov, K., Wiesner-Hanks, T., Barba, P., Cadle Davidson, L.E., Reisch, B. 2017. Computer Vision for High-throughput Quantitative Phenotyping: A Case Study of Grapevine Downy Mildew Sporulation and Leaf Trichomes. Phytopathology. https://doi.org/10.1094/PHYTO-04-17-0137-R.
Interpretive Summary: Measuring downy mildew growth is a common need in plant breeding and plant pathology. Observing a large number of samples can be advantageous, but requires time and money. We present a new approach where a computer program analyzes pictures to quantify downy mildew growth. The approach was tested on pictures of grape families having resistant and susceptible full-siblings. One family had shiny leaves, and the other had hairy leaves. In both families, a computer program (using pictures) and a researcher (using human vision) measured a similar amount of downy mildew growth. Additionally, the computer measured how hairy the leaves were. An experienced researcher would save 90% of their time using computer vision compared to human vision. Thus, more samples can be measured to better understand downy mildew resistance and leaf hair density, and money and time can be saved. We anticipate this computer vision system will find applications in other diseases or traits where responses can be imaged with enough contrast from the background.
Technical Abstract: Quantitative phenotyping of downy mildew sporulation is frequently used in plant breeding and genetic studies, as well as in studies focused on pathogen biology, such as chemical efficacy trials. In these scenarios, phenotyping a large number of genotypes can be advantageous, but is often limited by time and cost. We present a novel computational pipeline dedicated to estimating the quantity of downy mildew sporulation from images of inoculated grapevine leaf discs in a manner that is time and cost efficient. The pipeline was tested on images from leaf disc assay experiments involving two F1 grapevine families, one that had glabrous leaves (V. rupestris B38 x ‘Horizon’ [RH]) and another that had leaf trichomes (‘Horizon’ x V. cinerea B9 [HC]). Correlations between computer vision and manual visual ratings reached 0.89 in the RH family and 0.59 in the HC family. Additionally, we were able to use the computer vision system prior to sporulation to detect leaf trichome density. We estimate that an experienced rater spends over 90% less time scoring sporulation using computer vision compared to the manual visual method. This will allow more treatments to be phenotyped to better understand the complex genetic architecture of downy mildew resistance and of leaf trichome density. We anticipate this computer vision system will find applications in other pathosystems or traits where responses can be imaged with sufficient contrast from the background.