Abstract
Accurate segmentation of medical images is very essential in medical applications. We proposed a new method, based on combination of Learning Vector Quantization (LVQ), FCM and user interaction to make segmentation more robust against inequality of content with semantic, low contrast, in homogeneity and noise. In the postulated method, noise is decreased using Stationary wavelet Transform (SWT); input image is clustered using FCM to the n clusters where n is the number of target classes, afterwards, user selects some of the clusters to be partitioned again; each user selected cluster is clustered to two sub clusters using FCM. This process continues until user to be satisfied. Then, user selects clusters for each target class; user selected clusters are used to train LVQ. After training LVQ, image pixels are clustered by LVQ. Segmentation of simulated and real images is demonstrated to show effectiveness of new method.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have