Knife-Edge Scanning Microscopy: High-Throughput Imaging and Analysis of Massive Volumes of Biological Microstructures - High-Throughput Image Reconstruction and Analysis - page 21

Thus, Nissl staining allows us to reconstruct the distribution of all cell bodies in
the mouse brain, and in particular their distribution within the six layers of the
cerebral cortex. Golgi staining, in contrast, reveals the entire structure of neurons,
as it stains just 1% of the neurons in the tissue. Individual neurons can be seen
clearly, permitting reconstruction. India ink enables high-contrast staining of the
entire vascular network. Figure 2.6 shows a rendering of the data volume using
Amira [24].
2.3 Tracing in 2D
As we have seen in the previous section, the amount of data generated by high-
throughput microscopy instruments is staggering. Simply iterating through the en-
tire data set can take a long time. Thus, traditional image processing and 3D
reconstruction algorithms are not suitable for this kind of data, since they require
intense computation on every pixel/voxel in the data volume. Can et al. [21] and
Haris et al. [25] developed template-based methods to overcome this issue.
Figure 2.7 Identifying cross-section of fiber object and the moving window. An illustration of the
method used to identify fiber segments (the ''Y''-shaped object in the background) that overlap
with the circumference of the moving-window template is shown (the blacksolid pixels). First, all
the pixels are taken from the moving window's boundary, and convolution is done with a Gaussian
filter to obtain an intensity profile (bottom). The valleys in the intensity profile are identified as
fiber cross-sections. This process helps deal with inherent noise in biomedical images. Note that
the actual algorithm extracts a ''band'' of pixels from the window border, and applies a small 2D
Gaussian filter to extract the intensity profile.