THE APPLICATION OF FUZZY NEURAL NETWORKS IN THE SEGMENTATION OF HEAD MRI

The result of head MRIs segmentation is vulnerable to the edge blur and noise caused by the anatomical structure of head and the imaging process. By combining neural networks with fuzzy logic techinique, we presented a semi-automated method of segmentation of multispectral head MRIs based fuzzy neural networks(FNN). Experimental results showed that FNN was three times faster than the backprogagation neural networks(BP) under the same condition, and was more robust against edge blur and noise than maximum likelihood method(MLM), fuzzy c-means cluster(FCM) and BP.