Biomedical Engineering Reference
In-Depth Information
Fig. 2 Fitted coarse (a) and refined (b) individualised pelvic floor muscle models
subject. Mathematical landmarks that identified the largest PF muscle curvatures
and attachment points were used for this purpose, as they identified the dominant
features of the PF muscles. The definition of landmarks was verified by an expert to
ensure reliability and repeatability.
2.2 Model Geometrical Fitting
The affinely transformed PF models required further geometrical fitting to the
surface data to capture the local variations in the PF muscle surfaces. The
discretisation of the meshes was reduced before the first fitting of the geometry,
in order to avoid non-uniform spread of nodes and retain correspondence of
anatomical features. The fitted bicubic Hermite-linear Lagrange coarse models
reasonably approximated the muscle size and regional curvatures (Fig.
2a
). How-
ever, re-discretisation was required to generate models that adequately described
the detailed surface variations of the muscles. The second geometric fit was
performed using the refined trilinear mesh derived from the coarse fitted mesh.
Bicubic Hermite-linear Lagrange PF models were produced with uniformly spaced
nodes and without violating the anatomical feature correspondence required for
statistical shape analysis (Fig.
2b
). The resultant finite element models described
the PF muscle surfaces with reasonable accuracy while maintaining the same DOF
as the initial individualised PF models.
