Information Technology Reference
In-Depth Information
By computerized least-squares aligning of a parametric sine function to the
stereographically reconstructed landmarks, Stokes et al. [
74
] measured the Cobb
angle between the normals to the obtained curve at in
ection points in the coronal
and sagittal plane, and in the plane of maximal curvature. Drerup and Hierholzer
[
13
fl
15
] also considered the sine function appropriate, as it most resembled the
appearance of curves in idiopathic scoliosis. On the other hand, Patwardhan et al.
[
55
] justi
-
ed the use of spline functions by stating that splines are used to describe
geometries with continuously changing curvature, such as scoliotic spines. In their
framework for spine segmentation from CT images, Kaminsky et al. [
33
] used
spline functions because they proved appropriate to describe both the anatomical
shape and scoliotic deformations of the spine. Berthonnaud and Dimnet [
6
] con-
structed the spine curve separately in coronal and sagittal projections by computing
the average of two spline functions that connected the anatomical landmarks on
vertebral body walls. On the other hand, Peng et al. [
56
] used polynomial functions
to detect and segment vertebrae from MR images using vertebral disc templates.
Polynomial functions were also used to model both normal and pathological spine
curves in CT images by Vrtovec et al. [
83
]. The spine curve was automatically
determined by aligning the polynomial function with the centers of vertebral bodies
in 3D, obtained by maximizing the distance from the edges of vertebral bodies. The
same authors also developed a method for MR images [
86
], where the center of
vertebral body was first automatically detected in each axial cross-section by
maximizing the entropy of image intensities inside a circular region, and the
detected centers of vertebral bodies in 3D were then joined by a polynomial
function using the robust least-trimmed-squares regression. The method was also
used by Neubert et al. [
49
] for extracting the spine curve from MR images of high
resolution, obtained by applying the sequence named sampling perfection with
application optimized contrasts using different
fl
flip angle evolution (SPACE). The
work was continued by
tern et al. [
78
], who proposed a modality-independent
method for the determination of the spine curve that was extracted from locations
where lines connecting opposite edge points on vertebral body walls in the direction
of corresponding image intensity gradients most often intersected. Kadoury et al.
[
30
,
31
] determined the spine curve of a scoliotic spine in biplanar radiographs by
Š
first embedding cubic B-spline functions onto a non-linear manifold to predict an
initial curve according to a given database of scoliotic curves, and then performing
analytical regression to obtain a statistical model of the
final curve.
To extract the spinal canal centerline from CT images, Yao et al. [
96
] applied a
watershed algorithm followed by a graph search, while Hay et al. [
24
] applied the
fast marching minimal path technique that was based on the distance transform of
the spinal canal segmentation, obtained by morphological region growing. On the
other hand, Klinder et al. [
36
] segmented the spinal canal by a progressive adap-
tation of small tubular segments, represented as triangulated surface meshes, and
then determined the spinal canal centerline by calculating the centers of mass for all
contours of the obtained tubular mesh. A similar approach was proposed by
Forsberg et al. [
17
,
18
], who extracted the spinal canal centerline in CT images by
