Steps in the neuronal analysis of x-y positions.
The analysis is performed on individual pictures taken with a confocal microscope. For example, a picture from slice 10, SM series, column 1, row 3 is given by:
First, a wavelet filter is applied in order to smooth out the pixelized nature of the image and thus get rid of the high frequency noise, thus giving:
In order to identify individual neurons, a threshold is applied to get rid of all the background that do not contribute to neuronal photoemission. After this filter the picture is encoded in pixel values of 0 and 1 only, in contrast to the continous grayscale starting values from the original picture. The threshold value is chosen such that the most neurons appear as independent (not touching) objects, as compared to the original picture. The picture on the left has a threshold of 140 and the one on the right a threshold of 212, with respect to a 0-255 scale or gray color.
As a final processing, independent islands of black pixels are identified and sorted by their total size, or number of pixels. A lower size cutoff is then applied to the picture that takes care of remaining noise. Again, the size of the cutoff is chosen such that the smallest size object in the picture represents a real neuron, as compared with the original picture. On the left picture a cutoff of 40 pixels has been applied, while to the one right a cutoff of 20 pixels was applied.
In the final processed picture disconnected black areas are identified with neuronal bodies and their x and y coordinates of the center of masses are recorded. Although we performed the analysis by choosing two different thresholds and lower cutoffs, the positions of the x-y coordinates for both resulting pictures are not substantially different. The data set of the picture on the left (lower threshold, higher cutoff) is compared with the data set of the other choice of parameters (higher threshold, lower cutoff size).