By default SparkSpotter will analyse each pixel of your movie individually. If you're looking for a concerted effect in a larger area though you can extend the area over which SparkSpotter averages when analysing each pixel.
The value you set here is the number of pixels each side of the pixel being analysed SparkSpotter should average over. Eg setting it to 1 would make it use a 3x3 grid (with the current pixel at the centre), setting it to 2 would use a 5x5 grid etc.
This parameter is used in the filtering process to decide whether a spike is classified as an event. In the normalised dataset SparkSpotter calculates the mean and standard deviation of the values it collects. This parameter sets how many standard deviations above the mean a peak must be. This is useful as a high value in a noisy dataset might be of less interest than a similar value in a flat dataset containing a single peak.
The difference cutoff is the other filtering parameter you can use when selecting spark events. This sets an absolute difference in intensity units which a peak must be above the mean in order to be kept. This can be useful to eliminate very small peaks in pixels with very low noise (especially background pixels) which would have a low standard deviation and might therefore pass the SD Fold Cutoff.
The window size parameter is used in the normalisation (smoothing) step of the analysis. It says how may frames either side of the current frame will be averaged to calcluate the average against which your data will be normalised. Increasing this value will generate a more general, smoother curve through your data which will emphasise more long-lasting events. A smaller value will more closeley follow the data and will be better for identifying shorter events.