Biomedical Engineering Reference
In-Depth Information
in Fig.
1
a: both points were identified through auxiliary CMR short-axis planes and
LV axis was finally arranged after checking the operation on a supplementary LV
long-axis plane. For each acquisition, a set of 540 DICOM images were generated
and a total acquisition time of about 10min was required.
Offline CMR images segmentation was performed using a dedicated software
[
12
] developed in MATLAB (The MathWorks Inc., Natick, Mass). As detailed in
Fig.
1
c, two points (in red color) were manually selected on MV annulus in each of
the acquired plane, two points (in yellow color) were positioned on the free margin of
the MV leaflets and each papillary muscle (PM), where visible, was added through a
variable number of points (green points). Three-dimensional coordinates were then
computed using the information stored in the appropriate DICOM fields.
Mitral annular geometry was automatically generated through a 4th order Fourier
approximating function fitting the selected annular points; both anterolateral and pos-
teromedial PMs were positioned in the 3D space applying a
K
-means partitioning
algorithm to separate reference points in two mutually exclusive clusters and iden-
tifying the centroid of each PM tip. An additional set of points, was positioned on
the long-axis profile of each MV leaflet in order to derive MV leaflet profile through
a polynomial fitting (green lines), its local inclination (with respect to acquisition
axis) and leaflet extension, i.e. the distance between annulus and leaflet free-margin.
The segmentation of CMR images provided MV end-diastolic configuration and
the kinematic boundary conditions representing annular and PMs dynamics. In addi-
tion, MV leaflet surface was assessable at systolic peak, i.e. the mid-frame of systolic
interval within the set of CMR-phases, in order to reproduce MV morphology and
the severity of leaflet dysfunction through the extraction of different morphologi-
cal variables (e.g. leaflet billowing height and volume, regurgitation area, prolapse
extension, etc.).
2.2 Finite Element MV Pre-operative Model
Structural patient-specific FE models were derived from the CMR-extracted mor-
phology and integrated with intra-operative findings, in order to reliably reproduce
the specific MVP pattern; data from the literature were used to describe the mechan-
ical properties of leaflets, chordae tendineae and, in particular, artificial ePTFE
sutures. Once a pre-operative model (with chordal rupture) was simulated, the poten-
tial outcomes of neochordal implantation (NCI) were assessed, evaluating different
suture configurations. All simulations were carried out using the commercial solver
ABAQUS Explicit 6.10 (SIMULIA, Dassault Systèmes, Providence, RI).
MV leaflets were characterized by a single anterior segment and the peculiar
division of the posterior leaflet in three different scallops (i.e. P1, P2 and P3, respec-
tively): the end-diastolic geometry was discretized with a homemade software writ-
ten in Python (Python Software Foundation, Beaverton, OR): the mesh consisted
of 3-nodes shell elements (ABAQUS S3R element type) and a mapped scheme of
meshing was adopted. The number and the distribution of chordae tendineae were
