Four postmortem human brain specimens used in this study were obtained from The University of Iowa Deeded Body Program.

Histological Tissue Processing: The brains were perfused with 0.1M phosphate buffer followed by 4% paraformaldehyde in the same buffer. Thalami with adjacent parts of midbrain and basal ganglia from both hemispheres were dissected out and postfixed in fresh fixative for a period of one to two years. For establishing fiducial marks in the tissue blocks a hypodermic needle with a rounded sharp tip was driven through the anterior and posterior commissuers as they appear in the midsagittal cut. Sagittal sections were cut on a freezing microtome at 50 µm thickness in the plane parallel to the midsagittal plane. Each section was placed in an individual numbered compartment to maintain their sequence. Entire section series were mounted on slides, stained with thionin and used for analysis of cytoarchitecture.
Sections from the tissue block with 23 mm length of intercommissural line were chosen for imaging and 3D reconstruction. This length was comparable to that of the brain represented in the Schaltenbrand and Bailey atlas (23.5mm) thus facilitating comparisons of coordinates of different structures in the two datasets.

The choice of the sagittal section plane was determined by following considerations:
(i) it is easier to maintain the consistency of cutting angle when sectioning in sagittal plane as compared to coronal and horizontal, this in turn allows more acurate comparisons of different brains;
(ii) according to neuroanatomical data the projection zones of major subcortical afferent systems are arranged in antero-posterior sequence, hence their topographical relationships are better appreciated in this plane;
(iii) landmarks of the coordinate system, i.e., positions of anterior and posterior commissures in the midsagittal plane, can be easily marked and their projections easily followed in sagittal sections;
(iv) total number of sagittal sections is significantly smaller than that of coronal or horizontal sections.

Image analysis and 3D reconstruction: Nissl-stained sections were photographed with digital camera. Resulting digital images were processed in Photoshop correcting the brightness and contrast and removing small specks of dust and debri. Nuclear outlines were identified in an enlarged image of one section from each group of five, and verified under microscope comparing with staining patterns of immunocytochemical markers determined in an earlier study (Kultas-Ilinsky, 2011). Then the outlines of sections and nuclei were redrawn in Adobe Illustrator (1) and aligned (2) using fiducial marks. The drawings were then transferred to Image J, stacked, and the outlined structures were colored.

Further processing was done with Amira software: stacks were loaded as a 3D volume in the physical space calibrated in millimeters based on the 250µm section thickness and the pixel size determined by the original length of intercommissural line. Each structure was segmented by successive 3D thresholding of each color, smoothed, and converted to a 3D surface (3,4). The volume composed of all surfaces was rotated to align it with cartesian coordinate planes, and graduations were placed on the axes. The 3D volume was then cut in three different planes to create images of sagittal, coronal, and horizontal sections.

To illustrate histology images of each group of five adjacent 50µm thick sections was merged in Photoshop and edited using tools for enhancing brightness and contrast to bring out nuclear outlines.

For reconstruction of 3D version and registration to MNI space, as well as detailed description of the coordinate system and discussion of nomenclature used in the atlas plates see Ilinsky et al. 2018.