As a crucial procedure in which images are divided into distinct non-overlapping regions and the interested objects are extracted in the process of image analyzing and image recognizing, image segmentation plays a considerable role in extracting medical lesions, measuring specific tissues, and realizing three-dimensional reconstruction. In this paper, integrated with the studies of dental micro-CT images, an image segmentation algorithm based on textural features and neural network is presented to be applied in separating the targeted images from background. By textural-property analysis, the algorithm makes it possible for obtaining image characteristic parameter, which serves as input vector for neural network to extract characteristic information from targeted areas. It is presented in experiments that a relatively better segmentation effect can be expected to improve the accuracy and accelerate segmenting speed.