Background Flow diverter (FD) intervention is an emerging endovascular technique for treating intracranial aneurysms. High flow-diversion efficiency is desired to accelerate thrombotic occlusion inside the aneurysm; however, the risk of post-stenting stenosis in the parent artery is posed when flow-diversion efficiency is pursued by simply decreasing device porosity. For improving the prognosis of FD intervention, we develop an optimization method for the design of patient-specific FD devices that maintain high levels of porosity Methods An automated structure optimization method for FDs with helix-like wires was developed by applying a combination of lattice Boltzmann fluid simulation and simulated annealing procedure. Employing intra-aneurysmal average velocity as the objective function, the proposed method tailored the wire structure of an FD to a given vascular geometry by rearranging the starting phase of the helix wires. Results FD optimization was applied to two idealized (S and C) vascular models and one realistic (R) model. Without altering the original device porosity of 80%, the flow-reduction rates of optimized FDs were improved by 5, 2, and 28% for the S, C, and R models, respectively. Furthermore, the aneurysmal flow patterns after optimization exhibited marked alterations. We confirmed that the disruption of bundle of inflow is of great help in blocking aneurysmal inflow. Axial displacement tests suggested that the optimal FD implanted in the R model possesses good robustness to tolerate uncertain axial positioning errors. Conclusions The optimization method developed in this study can be used to identify the FD wire structure with the optimal flow-diversion efficiency. For a given vascular geometry, custom-designed FD structure can maximally reduce the aneurysmal inflow with its porosity maintained at a high level, thereby lowering the risk of post-stenting stenosis. This method facilitates the study of patient-specific designs for FD devices.