Cell Morphometry ZIP content by George McNamara
Robert Murphy's TypIC (typical cell Chooser,
now superseded by PSLID and SLIF
http://murphylab.web.cmu.edu/services/ ), was a "game changer" for me with respect to cell shape analysis. Rather than trying to compute the average of (say) a triangle and a pentagon ... which might result in a square, or a rectange, or some bizarre quadrilateral ... R.M. advocated using the median. OK, in a 2 member dataset this would result in averaging the two shapes (if use the standard way to calculate median of even number datasets), but this could be avoided in large datasets by stopping after acquiring an odd number of objects (cells), for example, N=101 (or N=99) instead of N=100 (if one happens to have N=100, could either pick a random number, and toss out the 'lower', or change the median of even number rule to be "pick one").
This is not to say the Elliptical Fourier Analysis (EFA) in the zip file is of no use: it is useful. In fact, a fun project (for someone else) would be to generate different triangle and pentagon pairs, compute their EFA harmonics, average their harmonics, and inverse Fourier transform back to a real shape. This could be extended to other pairs or sets of shapes. There is plenty of cell shape data online, more EFA could be done with those.
The zip file also includes information on Temporal Area Maps (TAM) and TAM histograms. See
for my "Rogers PMN Movie Panorama 2012 The Chase".
This zip file includes a text file - and other content - explaining how to generate Temporal Area Maps in MetaMorph Imaging System (versions 1.0 through 7.7 ... Stack Arithmetic: Sum, should also be in MetaMorph NX).
Request to ImageJ developers: Please implement the equivalent of "Stack Arithmetic" (sum, average, median, minimum, maximum, best focus, AND, OR, XOR) in ImageJ. After writing the email message that turned into the text file I looked in Fiji ImageJ for something like "Stack Arithmetic" and was unable to find anything useful. Considering how long ImageJ has been around, I am amazed at how clueless the ImageJ developer community has been about simple but powerful image processing. I did the first Temporal Area Map by hand in Robert P. Futrelle's lab (we were at UIUC, circa 1984). Bob implemented TAM in Lisp and C on a VAX 11/730 from cell outlines acquired using the setup in Futrelle et al 1982 J Cell Biology http://jcb.rupress.org/content/92/3/807 - free online access). I later implemented TAM in Image-1/AT (circa 1990, Neal L. First's lab, UW-Madison, my thanks to David Szent-Gyorgyi and others at UIC for help controlling the Matrox MVP-AT board with Turbo Pascal ... this simplified TAM compared to my initial Image-1/AT journal). Very soon after I joined UIC in 1992 the MetaMorph developers tweaked Stack Arithmetic Sum to make TAM trivial with a binary stack (summer 1992, pre-version 1.0 release in December 1992). Upshot: 20+ years later (I am writing this in March 2014), the ImageJ community is still clueless.
- cell shape,
- Elliptical Foutier Analysis,
- Temporal Area Maps,
Available at: http://works.bepress.com/gmcnamara/41/