Wang, Liya , Spalding, Edgar , Assadi, Amir .
A computational method for automatic quantification of seedling photomorphogenesis phenotypes based on principal components analysis.
Growth and curvature of Arabidopsis hypocotyls can now be automatically quantified by a new computational algorithm that operates on time series of electronic images of seedlings responding to light. The pixels comprising the seedling stem (subject) are isolated from the background with an adaptive threshold algorithm and then transformed into a three dimensional vector consisting of x, y, and intensity values. A local center of mass (LCM) is determined within an analysis window that is then moved by an adjustable step size in a direction determined by local principal component analysis (LPCA). The LCM within the window centered at new location is determined and LPCA again determines the direction in which the analysis window moves by one step. A weighting function that incorporated an intensity function and a Gaussian distribution is used to ensure that the midline is not misdirected in regions where the folded cotyledons touch the hypocotyl or where the subject was significantly curved within the length of the analysis window. The coordinates of LCMs determined by this iterative, LPCA-driven process form an ordered set of midline points that capture the length and shape of the subject for each image in the time series. A reliable intensity difference feature determined the position of the hypocotyl/cotyledon boundary along this midline in each image so growth rate, axial curvature distribution, and apical hook angle of the hypocotyl can be calculated without interference from the cotyledons. Examples of this algorithm being used to quantify features of mutant phenotypes will be presented.
Log in to add this item to your schedule
1 - University of Wisconsin, Department of Botany, 430 Lincoln Drive, Madison, WI, 53706
2 - University of Wisconsin, Department of Botany, 430 Lincoln Drive, Madison, WI, 53706
3 - University of Wisconsin, Department of Mathematics
Presentation Type: Plant Biology Abstract
Location: Exhibit Hall (Northeast, Southwest & Southeast)/Hilton
Date: Sunday, July 8th, 2007
Time: 8:00 AM