Figure 4. Algorithm for locating a soma in a local sub-stack.

(A) The original image sub-stack. (B) The maximum intensity projection image of the sub-stack in (A) along three orthogonal orientations. (C) The images after Gaussian smoothing. (D) The images after grayscale morphological closing. (E) The binary volume intersection including the soma volume achieved by the backprojection of the three projection images into 3D image space. (F) The images including the soma centroid obtained by connected component analysis and computation of the center of mass.