To do this, I am using two different programs, Multispec and ImageJ. Multispec is designed for classifying satellite images while ImageJ has a wide variety of different applications, including measuring the area of photos.
While I have not done this using my samples, I have been analyzing eelgrass shoots for UW Friday Harbor Laboratories. Earlier in the week, Dr. Yang and I completed a transect of an eelgrass meadow in Ship Harbor Park. We collected the 20 longest unbroken leaves and took them back to SPMC for processing. After they were scanned into the computer, I used Multispec to teach the computer which parts of the picture were healthy eelgrass and which parts were infected with Labyrinthula zosterae. I also had to train the computer to recognize the background and other kinds of damage. After the training was done, a new grey-scale image was created. Each color was associated with a different class (healthy, infected, damaged, or background). Then I could use ImageJ to measure the areas of different colors and get the precise area of damaged and healthy tissue.
This technique will be helpful in my experiment in order to quantify the damage caused by herbivores and later, the disease. One of my hypotheses is that the exposed inner tissue of the eelgrass caused by herbivore eating will increase the speed and likelihood of the infection because Labyrinthula sp. spreads by direct contact. This process will give me a less biased way of comparing herbivore damage to infection damage in order to test my hypothesis.
|Original eelgrass shoots|