D. Measuring mask and measurements
Fig. 9: Right panel: Image monitor with outline of the measurement area in one field of view (yellow lines) with highlighted horizontal measuring mask. Left panel: Notation for numerical estimate of basic morphometric parameters, corrected by measured value of excluded image elements (“artifacts”, vessels etc.). See explanation in the text.
Three morphometric parameter were measured: area fraction AA, numerical density NA, and the mean cell area SA (see Fig. 9). To increase the vertical resolution of the obtained curves, which will characterize the cortical profile according to the selected parameters, the special measuring procedure was performed. The scanning mask Z(scan) was formed as the thin horizontal strip of vertical width 35 microns which was approximately equal to the vertical diameter of the average neuron. This mask was moved during the measurements along the vertical direction of neuronal unit. The morphometric values were stored in different arrays according to the depth of the current field of view and the scanning mask position within the field. The images of all particles, eliminated from the measuring field, were accumulated in the separate bit plane. The measuring mask was prepared as the complement to the union of all eliminated particles in the current field (Z mes., Fig. 9). The measuring values were estimated as a fraction of the current parameter and the area of scanning mask, intersected with the measuring field. This approach reduced the underestimation of the measured morphometric parameter in the field with artifacts and vessels. The image particles, touching the upper and the left edge of the measuring mask were eliminated before the measurement. The area fraction was measured from the image, prepared by transformations described above. For other parameters, the image was additionally transformed according to the following procedure. The current value of cells number was calculated as the number of geodesic centers, which intersect the scanning mask, where centers of cells were found by ultimate erosion of X4 (see Fig. 7, notation on the right panel). This allows to attribute the position of every neuron to the given depth according to the coordinate of their center. Also, the repeated measurement of the neurons was avoided which belong to several scanning masks together. The estimation of the mean area of a cell SA was obtained as a fraction of the area of cells Xr and centers number. The set Xr was reconstructed from the centers of binary blobs, which belong to the scanning mask.
Finally, we would like to demonstrate some performance data of described algorithm. For technology developed almost 40 years ago it is not bad at all, is it?
Cortical profiles, obtained using described technology, were used in three different areas: comparison of cortical architecture in some human brain areas in normal conditions, in studies of cortical ontogenesis, and in comparison of architectural differences in normal condition and schizophrenia.