An Encoder-Segmented Neural Network (ESNN)-based approach is proposed to improve the efficiency of image segmentation. The features are ranked according to the encoder indicators by which the insignificant features will be eliminated from the original feature vectors and the important features reorganized as the encoded feature vectors for the subsequent clustering. The ESNN developed can improve on the existing Fuzzy c-Means (FCM) algorithm in feature extraction. The cluster number selection can be accomplished automatically. This method was successfully implemented for automatic labeling of tissues in MR brain images. Experimental results show that the ESNN-based approach offers satisfactory performance in both efficiency and adaptability.