Biomedical Engineering Reference
In-Depth Information
assist with further imaging or for augmenting biopsy or surgical procedures. The
effectiveness of these methods largely depends on how well they can represent the
mechanical response of the different breast tissues under these various loading
conditions.
Previous studies have simulated individual-specific deformation from the supine
position (typical for surgical procedures) to the prone position (typically used in
MR scanning) using homogeneous finite element (FE) models of the breast [
1
].
However, these studies required information derived from both supine and prone
MR scans, and applied observed displacement boundary constraints to the models,
which diminishes their capacity to predict deformation based only on the loading
conditions. Other studies have accounted for this by modelling the supine to
standing position using heterogeneous FE models constructed from supine CT
scans for the purposes of surgical planning [
2
]. During clinical imaging, however,
prone MR scans are most commonly acquired; therefore, developing a framework
for predicting deformation from the prone position to, for example, the standing or
supine position would be beneficial.
The present study builds on previous 3D subject-specific homogeneous models
created for predicting deformation from the prone to supine positions [
3
] and aims
to extend the outcomes to include different mechanical responses for the different
breast tissues, in particular the pectoral muscles, which have previously been
neglected in studies involving gravity loading [
1
,
2
]. The region of tissue above
the pectoral muscles in the upper outer or shoulder region of the breast is important
since tumours are most commonly located in this area [
4
], as identified through
lymphoscintigraphy [
5
]. It is therefore critical, from a clinical perspective that
breast biomechanical models accurately predict the motion of tissue in this region
of the breast. Modelling muscle deformation within the breast is also important
because of the high ratio of stiffness between muscle and both fat or fibroglanduar
tissues, which would influence the biomechanical response of the breast.
To demonstrate the applicability of these heterogeneous models, we used segmented
data from reference supine MR images to provide information for identifying the
mechanical properties of an individual's breast. The results from these simulations
were compared to those obtained from homogeneous models in order to determine the
importance of including the muscle in biomechanical models of the breast.
2 Subject-Specific Breast Geometry
Subject-specific FE models of a volunteer's left breast were created by fitting cubic
Hermite shape functions [
6
] to skin and rib-muscle surface data segmented from
prone MR images [
3
] as shown in Fig.
1
. The skin and rib-muscle root-mean-
squared errors (RMSE) following the fitting were both 0.8 mm. Segmentation of the
internal breast tissues into the constituents (fat, fibroglandular and muscle) was
performed in a semi-automatic manner, using intensity thresholding, manual
editing, and connected component analysis [
7
].
