Multi-Objective Design of a Biodegradable Coronary Artery Stent - Cardiovascular and Cardiac Therapeutic Devices - page 21

Fig. 16 Response surface model for non-dimensional inverse of minimum lumen area. The
abscissa on each small tile represents non-dimensional strut width. The ordinate on each small tile
denotes a non-dimensional parameter related to the shape of the crowns and each tile signifies
fixed values [0.00, 0.25, 0.50, 0.75, 1.00] of the third and fourth parameters (which cannot be
disclosed due to intellectual property protection)
So, the multi-objective problem comprised four input geometry parameters and
four output objectives. In the time available for simulation, it was decided to run a 40
point design sample to gain an understanding of the multi-objective design space.
Some of this data resulted in the response surface models shown in Figs. 16 , 17 .
The response surface model for the inverse of MLA is shown in Fig. 16 . The
inverse is used so as to treat each objective as a minimization problem. Each tile
represents a slice through the non-dimensional 1/MLA design space for fixed
values [0.00, 0.25, 0.50, 0.75, 1.00] of the other two non-dimensional geometry
parameters. Since all tiles are very similar, these two parameters have little effect
on MLA. However, within each tile, favourable values of MLA are observed for
high values of strut width (shown on the abscissa) and for low values of the
ordinate. This follows a similar pattern to that shown in Fig. 12 and, again, wider
struts predict greater radial support.
Also echoing the findings in [ 25 ], volume averaged tissue stress (VAS) exhibits
a response that is in competition with MLA as depicted in Fig. 17 . It is desirable to
minimize tissue stress and, based on the definition of VAS [ 21 ], low stress occurs
for low strut width, due to the lower stent-to-artery ratio of narrow struts and the
associated reduced load needed to support the higher recoil in the vessel.
Combined in a multi-objective plot, the relationship between MLA and tissue
stress can be more readily interpreted as in Fig. 18 , wherein each point signifies an