Study on Positioning Center of Fascicular Groups in Nerve Slice Image

Acquiring positioning centers and number from the nerve slice images automatically is one of the key to extracting the edges of fascicular groups. Essentially that is to cluster those pixels in the nerve slice image which represent the fascicular groups. Because those pixels are distributed discretely and the fascicular groups often split and converge quickly while extending, this paper designs the aim function in order to get the optimum fascicular groups number, which is the sum of the inner-class mean distances and the mean distances among classes. Then the k-means method is employed to cluster those pixels which represent the fascicular groups and the fascicular groups´ number is enumerated from 5 to 30 and corresponding aim function values are calculated. Subsequently the minimum aim function is obtained after being compared, which is corresponding to the optimum fascicular groups´ number. The experiments show that the right positioning center and optimum number of fascicular groups from the slice image can not only be obtained automatically but also calculated very quickly through the above mentioned method, which is the important basic of 3D visualization of nerves.