In　order　to　reduce　the　risk　of　intracranial　aneurysm　rupture　after　implantation　of　stent　with　trapezoidal　cross-section　wire,　attentions　are　paid　to　the　optimization　of　cross-section　of　stent　wire.　38　models　of　different　extents　of　baseline　of　stent　with　trapezoidal　cross-section　were　created　in　SolidWorks　2008.　Then　ANSYS　12.0　with　fluid-solid　inter　action　method　and　Generalized　Regression　Neural　Network　(GRNN)　were　used　to　map　the　nonlinear　relationship　between　the　maximal　pressure　gradient　on　aneurysm　wall　and　the　extent　of　baseline　of　the　trapezoidal　cross-section　of　stent　wire.　Genetic　Algor　ithm　(GA)　was　employed　to　obtain　the　optimum　value　of　baseline　of　the　stent　by　minimizing　the　maximal　pressure　gradient.　The　results　indicate　that　the　maximal　pressure　gradient　was　reduced　by　7.86%　after　optimization　compared　with　the　traditional　stent　with　rectangular　cross-section　wire.　The　combination　method　of　GRNN　and　GA　was　effective　approach　for　stent　optimization.　©　2013　Springer-Verlag.