|
|
Hello,
I've been using ImageJ for quantitative analysis of western blot data. For
the past two years, I've use the Rodbard equation to calibrate ImageJ to a
step tablet to convert gray values to optical densities. From there, I can
determine the optical densities of my samples and determine a rate of decay
of the sample of interest. I learned how to use ImageJ by following the
tutorial for gel analysis that is included with ImageJ.
Recently, I learned how to take a histogram of the region of interest. I
have also recently learned that the calibration curve converts a linear
scale, the gray value, to a logarithmic relationship, the optical density.
I have noticed that if I take a gray value and plug it into the Rodbard
equation using the values determined during calibration that I don't come up
with the same value for optical density that ImageJ reported based on the
calibration curve. For example, if the gray value of 1 corresponds to an
optical density of, say, 3.5, plugging the value "1" into the fitted Rodbard
equation gives me a large negative value for the optical density. I have
not made a simple mathematical error, and do not understand this
discrepancy.
I tried using a lograithmic fit. Using a logarithmic fit (y=a ln (bx)), I
was able to return a fitted value by plugging a gray value into the fitted
equation. For example, let's pretend the logarithmic fit said that a gray
value of 1 equaled an OD of 3.5. If I then plugged "1" into the fitted
equation, the result was 3.5 for the OD. So far so good. The problem I
have is this. Both the Rodbard and the logarithmic fits appear to generate
curves that fit the calibration data very nicely. If I use the logarithmic
fit to analyze my data, in the end I have rates that are twice as long as
those determined using a step-tablet calibrated with the rodbard equation.
In short, I am confused about what ImageJ is doing and seek help. I hope to
find someone with significant skill using ImageJ for quantitative analysis
of black and white images, especially western blots.
Thanks,
Mark
|
Locked
