Acquisition of Stem Images:
The prepared slides of soybean Glycine max stems (Collected in 2011) were observed with a Nikon Eclipse Ti-E at 10x objective and .7 aperture. At a 10x objective, the entire cell was not able to be captured in one image, but instead multiple images with a 10% overlap of the cell. In addition, a .7 aperture was recommended because with a higher aperture, the more light the microscope can allow in the lens to bring in out more details such as the safranin color of the xylem). Therefore, coupled with a Nikon DS-Fi2 camera using NIS-Elements Advanced Research 4.5 software (Richmond, Virginia, United States) the soybean stems were acquired at 2560×1920 full resolution.
Stitching Images into a Composite
ImageJ-Fiji is a software with many plugins that can be applied to scientific image analysis. The plugin Grid/Collection Stitching (Plugins>Stitching>Grid/Collection Stitching) was used to take several images of a single cross-section and fuse it into one image. The Grid/Collection plugin uses the image file name (i.e. tile_x001_y001) and assigns each picture a location where each picture with 10% overlap each other to provide an accurate display of the cross-section through stitching several images together.
Segmentation of Soybean Images
(Images were normalized using…) The overall objective of soybean segmentation is to be able to segment all tissues automatically. ImageJ-Fiji recognizes patterns, pixel density, texture, and saturation and separates the image into different stem tissues; pith, xylem, phloem, phloem fibers, and cortex (Fig. 2). However, due to the limited amount of time, the focus of segmenting the xylem tissue would bring a promising method towards segmenting the rest of the tissues in the future. With the acquired image at full resolution, we created an automated set of commands (macros) that allowed us to begin the segmenting process.
Statistical/Data Analysis
Once the xylem tissue has been separated through the automated process, the macro will display the xylem area and the area of the soybean cross-section in pixels. The percent of xylem can be found by this formula: = Percent of xylem. This data was then compared to the data of the same cross-section obtained through a manual, hand traced method. The manual and automated data was analyzed by using three samples from the same branch and running it through a two tailed t-Test in Microsoft Excel. The t-Test shows whether or not the two methods are significant or not significant based on the P-value. Therefore, if the P-Value > .05, then the two methods is not significant, meaning there is no real differences between the two methods. This would support that the automated process is able to accurately measure the area of the xylem, efficiently acquire the data, and eliminate possible variables that may come from hand tracing for a long period of time (i.e. Fatigue, shaky hands, carpel tunnel)