I just thought I’d take advantage of this thread to mention a few issues with measurements of intensity involving a threshold on the same channel as the one being measured. And more specifically using automatic thresholds.
@Felix, you are measuring the intensity of the cardiac fibers in order to compare different conditions, I take it. If you select multiple areas by hand, avoiding the holes, and measure them, then this is fine.
However, if you start using an automatic threshold, as suggested by @stelfrich you run the risk of biasing your results in exactly the way that you do not want. In short, auto threshold methods select a threshold based on the distribution of intensities in your image. If the distribution changes, then the threshold will change too. But you should not change the threshold, seeing as your’re changing how you measure each image.
Just like having an actin lane in your western blots, you need something that does not change in your data to measure what is changing accurately. In your case a membrane or cytoplasmic marker that is not affected by your experiments can help you detect the cell area so you can measure your fibers.
I’ve attached a figure to explain what I mean by bias. Thresholds on the same channel as the measured ones will lessen the potential variations you can observe. In this example case, we have DAPI stained cells always present and a GFP signal that is dependent on cell cycle. If I use only the GFP signal, if I select a too high threshold, I lose some cells with low intensities and thus believe my measured signal is higher than it actually is. Converesely if my threshold is too loose, I risk taking in too much background and artificially lower my measured signal.
Using a mask based on the DAPI, I can measure all cells for the green in a much more unbiased way.
Anyway, TL;DR feel free to draw out the regions, or set a hard manual threshold based on negative controls to make sure you are capturing specific signal and not background fluorescence and stick to it. Do not use automated threshold methods to detect AND measure the same thing.
All the best