Fw: Which graphic?

[Jelena Vasiljevic]

Dear Audrey,
 
Thank you very much, but I still get several graphics for each of parametars, for example F vs Alpha, even with settings that you adviced me: select "Show location with highest CV" and "print slope and CV for Dq vs Q", unselect the omit box count checkbox before doing the scan. Do I have to choose handly the graph that has the highest CV (but what CV, there are several types)?
And is there any chance that I can scan the whole picture, not just one region?
Thank you very much!
Jelena
  ----- Original Message -----
  From: audrey karperien
  To: Jelena Vasiljevic
  Sent: Wednesday, February 22, 2006 7:42 AM
  Subject: Re: Which graphic?


  Hi.  I am having some trouble understanding what you need to know, but I hope I've got it right this time.  If you are doing a standard multifractal scan (not a random mass multifractal scan), then you can use the graph that has the highest CV.  The program will generate the data for only this graph, and the graphic itself, if you select "Show location with highest CV" and "print slope and cv for Dq vs Q".  To access the raw data for that scan, unselect the omit box count checkbox before doing the scan.
  Let me know if this is not what you needed.
  Audrey


  Jelena Vasiljevic <[hidden email]> wrote:
    Dear Audrey,



    Thank you very much for your thorough explanation. That was very useful, but I also need an information about next issue: which one of several graphics that I get for each parametar should I choose to put into further statistical analysys? For example, when I analize one picture (precisely, region of the picture), I get several graphics for F vs Alpha (the same case is for D vs Q, etc). Because I need only one graphic for each parametar, which one (and I need only one) of those graphics should I use. Should I, perhaps, use mean values of given extremes? Please, help me with this :)



    Sincerely,

    Jelena Andjelkovic











    ----- Original Message -----
    From: "audrey karperien"
    To:
    Sent: Friday, February 17, 2006 6:23 PM
    Subject: Re: SOS What graphic should I use?


    >I forgot to mention to include images of known mono-and multi-fractals and
    >euclidean forms of similar size and foreground pixels to compare to your
    >data images.
    >
    > audrey karperien wrote: Hello.
    > 1. It is likely that you can set it to 1 scan location. The number of
    > scans depends on the sampling you need. FracLac 1) randomly samples parts
    > of an image then reconstructs the distribution, and more effectively, 2)
    > samples the entire image multiple times after partitioning it in different
    > ways. Bottom line: 1 will probably be good, but if there is a big
    > difference in the graphs when you set it to 4, then you need to consider
    > more data.
    > 2. Use DQ vs Q plots graphics and for statistical analysis, use the slope
    > of DQ vs. Q and the coefficient of variation. The plot is usually either
    > essentially unchanging or decreasing. For simple monofractals, DQ is
    > usually essentially unchanged, with low dispersion, over Q so the DQ vs. Q
    > plot is stable and generally non-decreasing. For multifractals, in
    > contrast, DQ typically decreases with Q, being sigmoidal around Q=0.The
    > coefficient of variation is one way to measure the variation (CV =
    > standard deviation over the mean). The CV for the graph of DQ vs. Q is
    > generally greater for multifractals than monofractals.
    > 3. Also use f(alpha) vs alpha plots, converging = not multifractal;
    > broadly humped = multifractal. The graph of a monofractal for f(alpha) vs
    > alpha is typically converging (especially for Q>0) but for multifractals
    > f(alpha) rises and falls over a broad hump with alpha.
    > 4. Use the default minimum pixel ratio. That setting is important for
    > excluding nonrepresentative image samples.
    > 5. Strengthen your argument using sliding lacunarity which adds
    > information for understanding scaling of patterns in digital images:
    > multifractal patterns are characterized by negative Dq vs Q, with high
    > CVs, diverging humped f(alpha) spectra and a typically humped then
    > decreasing lacunarity plot; monofractals converge in the f(alpha) plot
    > (Q>0) and generally increase in lacunarity; non fractals converge over the
    > f(alpha) vs alpha plot and for simple forms usually show low, stable
    > lacunarity.
    >
    > Jelena Vasiljevic wrote:
    >
    > Hi
    >
    > I am doing my thesis on multifractal analysys of medical images, and need
    > any kind of help I can get. When I've analyzed images with ImageJ , for
    > every analized image I've got several graphics with different values of F
    > vs Alpha, Q, F vs Q, Alpha vs Q, etc. Should I use mean values of each of
    > those variables for statistical analysys, or the last one made graphic or
    > something else? I do that analysys on images of three groups of cells, and
    > my aim is to find some statisticaly important diferences between those
    > groups (regarding Alpha, Q, FvsAlpha...). I am not sure which values for
    > "Minimum pixel ratio to accept in samples" and "Number of global scan
    > positions" should I set. Could you please give me some suggestions, it is
    > very urgent?
    >
    > Thank you in advance!
    >
    > Sincerely,
    >
    > Jelena Andjelkovic