2
The negatives are scanned at 2700 dots per inch (dpi)
using a Nikon Coolscan–1000 35 mm fi lm scanner inter-
faced to a Macintosh PowerMac G3 computer. To bring
out all of the subtle shades of gray, the exposures are
scanned as color positives rather than as black and white
negatives. The software driving the scanner is a plug–in for
Adobe Photoshop and generates raw fi les that are approxi-
mately 27 megabytes (mb) each.
The raw fi les are reversed to positives and are
changed to grayscale (reduces fi le size to 8 to 9 mb). The
image resolution is set to 300 dpi. Using the enhancement
features built into Adobe Photoshop and the additional fea-
tures of Extensis Intellihance, adjustments are made to
bring out contrast and increase sharpness. A normal-con-
trast image is in part A of each companion plate along with
a low-contrast copy in part B. Labels are placed onto the
low-contrast image using Adobe Illustrator. Small copies
of each normal-contrast image are shown in the overview
plates. A small image showing the spinal cord with the
surrounding structures of the vertebral column is included
in the plates of the fi rst 11 specimens since the spinal cords
have not been dissected from the surrounding tissues. For
specimens in the late fi rst, second, and third trimesters, the
area surrounding the section is removed from the images.
D. Methods: 3D Reconstructions
Cervical Cord in the First Trimester (Part VIA):
Eight specimens were used to reconstruct the cervical part
of the spinal cord from GW 4.0 to GW10.5, following a
5-step procedure. First, photographs of sections are made
at regularly spaced intervals through the cervical spinal
cord; the negatives are scanned and converted to computer
fi les (as described in Section C above). Second, all the fi les
from a single specimen are placed into one large Photoshop
fi le that contains a separate photograph in each layer. By
altering the visibility and transparency of these layers the
sections are aligned to each other using the middle dorso-
ventral part of the spinal canal as the reference for align-
ment. In the living state, there is a curve in the cervical
region of the developing cord; that is diffi cult to reconstruct
accurately and often makes interpretation of the rendered
images confusing. Therefore to simplify the reconstruc-
tions, the centers of the sections are aligned straight. Third,
Adobe Illustrator is used to outline the following 14 struc-
tures in each section: (1) the outside edge of the entire sec-
tion, (2) the entire neuroepithelium on the left side, (3–5)
the dorsal, intermediate, and ventral parts of the neuroepi-
thelium on the right side, (6–7) the roof and fl oor plates, (8)
the undivided gray matter on the left side, (9–11) the dorsal
horn, intermediate gray, and ventral horn on the right side,
(12) the dorsal funiculus on both sides, (13) the lateral
funiculus on both sides, and (14) the ventral funiculus on
both sides and the ventral commissure. Since structures
are usually symmetrical in most specimens, the outlines
from one side are copied to the other side to simplify the
images. The outlines from each section are saved in sepa-
rate Adobe Illustrator eps (encapsulated postscript) fi les.
Fourth, the eps fi les are imported into 3D space (x, y, and z
coordinates) using Cinema4DXL (C4D, Maxon Computer,
Inc.), a modeling and animation software package. For
each section, points on the outlines have unique x–y coor-
dinates and share the same z coordinate. By calculating
the distance between sections, the entire array of outlines is
stretched out in the z axis. The outlines are segregated into
14 different goups, one for each structure. The C4D loft
tool builds a “skin” for each structure by creating a spline
mesh of polygons. The polygons start from the x– y points
on the fi rst outline with the most anterior z coordinate,
to the x– y points on the next outline behind it, and fi nish
with the x–y points on the last outline at the most posterior
z coordinate. The polygons are rendered either as com-
pletely opaque or as partly transparent surfaces using the
C4D ray-tracing engine. Selected parts of the model can
be made either invisible or visible during rendering using
the various options in C4D. Fifth, the rendered images are
converted to Adobe Photoshop fi les, and Adobe Illustrator
is used to draw thin lines on some of the surfaces to make
the images more easy to understand.
Motor Columns in the Ventral Horn (Part VIB):
Using the same steps as described in the preceding para-
graph, the entire spinal cord from upper cervical to sacral
levels is reconstructed in three specimens at the end of
the fi rst trimester and at the beginning of the second trimes-
ter (GW8.5, GW10.5, and GW14). The following struc-
tural outlines are drawn: (1) the outside edge of the entire
section, (2) the gray matter on both sides, (3) the central
canal, (4) motoneuron columns in the right ventral horn,
(5) motoneuron columns in the left ventral horn. The
images are rendered so that the white and gray matter are
shown as transparent, glass-like surfaces surrounding the
opaque surfaces of the central canal and the motoneuron
columns. Because the spinal cord is many times longer
than it is wide, the thin motoneuron columns are diffi cult to
see when all axes are rendered at the same scale, especially
in the GW14 model. Consequently, the x–y axis is three
times larger than the z axis in all models.
E. Terminology
The labels in this Atlas come from a review of the
literature (Altman and Bayer, 2001) on experimental stud-
ies of anatomical connections in the spinal cord of various
mammals. In addition, several atlases of the adult human
spinal cord are used (Fix,1987; Roberts et al., 1987; DeAr-
mond et al., 1989; Haines, 2000) and some classical neu-
roanatomy books are consulted (Ranson and Clark, 1959;
Crosby et al., 1962; Truex and Carpenter, 1969; Brodal,
1981). Some of these references use the directional terms
anterior and posterior. Since our orientation is to link the
work in humans to experimental work in animals, dorsal
and ventral are the directional terms used throughout this
Atlas. For the most part, the labels follow the terminol-
ogy in these atlases and texts, especially in the gray matter.
Introduction
