Medical Image Segmentation by using Active Flow Model with Simulated Annealing

This study suggested the active flow model with simulated annealing as the effective segmentation method for medical image. The technique proposed in this study emulated the previous active flow model by applying the simulated annealing to adjust the α parameters. The technique's success was examined through the samples of synthesis, CT, and ultrasound images. The samples were tested with three different texture segmentation models including the traditional active contour model, active flow model, and active flow model with simulated annealing. The results showed that the active flow model with simulated annealing achieved better segmentation on CT image, yet lower precision on synthesis and ultrasound images. Accordingly, the technique needs improvement in flexibility for α parameter adjustment.