High-resolution Nissl-stained sections of human brain were downloaded from Allen Brain Atlases portal using published API. Image resolution was 1 micrometer or better. Depending upon the section region, total size of JPEG files varied from 300MB to 1.5 GB. Between 10 to 15 extra-large fragments from each section (~5000 pixels wide, ~3000 pixels high) were selected on low-resolution images using proprietary viewer (“Qute Brain Viewer”, V.2) and saved with original resolution. Attribution to target cortical area in each section fragment was determined using available annotations.
Fig. 1: Example of initial images. Left- area 10m (4565×3658), right – area 37 (2659×2651).
Segmentation of each section was produced using watershed algorithm (“Segmentation using algorithms of Mathematical Morphology“). Total area of all processed section fragments for each area was close to 50 mm.sq.