Optimization of parameter for extracting brain regions using deformable contour models
Seiya Odagiri, Kazuhito Sato, Hirokazu Madokoro · Society of Instrument and Control Engineers of Japan · 2012
This paper presents a method that considers sharing of parameters of Level Set Methods and optimizes method using Genetic Algorithm. We optimized parameters of LSMs that is deformable model using GA of evolutional learning. However, the proposal method needs Ground Truth for each clinical image. Therefore, application with clinical images without GT was challenging task. In this paper, we focus on shaering of parameters and try application of the proposal method for clinical images without GT for an individualization trade-off problem. Moreover, the extraction accuracy performed the optimal value search for iterations of updating to the low case using Active Appearance Models after sharing.