This plug-in of ImageJ (NIH) measures the length, width, surface area, and the volume of a rod shaped cell or spherical spores given that the user draws the outline of the cell as an ROI.
This plugin works with ImageJ and it requires the JAMA package to carry out the matrix manipulation.
Download the .class file into the plugins folder of your ImageJ installation folder. Alternatively, download the source file to the plugins folder then compile and run. Also make sure that you have JAMA package in the same plugins folder. If you do not have JAMA, download it from NIST JAMA website. Start or restart ImageJ if you had it open.
Source file: Pombe_Measurer.java
Class file: Pombe_Measurer.class
Jama package: Jama-1.0.3.jar from NIST website
Draw the outline of the cell using polygon tools in ImageJ. The input of the plugin needs an area selection. Click Plug-in -- Pombe Measurer and it will pop up to let you input the pixel length. This window will only pop up once. If you need to change the pixel length later on, restart ImageJ. Once you have entered the pixel length, click OK.
When processing rod-shaped cellular images, one time-consuming task is to rotate the long axis of the cell in the image plane to be horizontal. In Pombe Measurer, the actual coordinates of the outline of the cell are used and Principle Component Analysis is performed to find the long axis of the cell. Then a rotation matrix is calculated from this angle to rotate the image. No manual work is needed after the plug-in launches.
The cell is then sliced vertically into thin slices (typically with one pixel width). The surface area and the total volume of the cell are approximated by aggregating the surface area and the volume of the cone frusta.
Because of resolution limit and the algorithm cannot approximate the surface area and volume measurements using thin slices under pixel resolution.
The worst performance is with perfect spheres in that approximation is least accurate. For example, a 100 pixel diameter sphere will have a 4.6% underestimation in surface area. When it is only 30 pixels in diameter, this underestimation is about 9%. This may be improved in the next version.
The surface area measurement of long rods is more accurate and the measurement of volume is accurate in general (error < 1%) with worst case performance in perfect small spheres of less than 5% underestimation.
All comments and suggestions welcome! Thank you in advance!