Segmentation of Melanoma and Lymphocyte Cells in Three-dimensional Collagen Matrices

Computer-automated detection of viable cells of varying size, shape, and contrast is a prerequisite for monitoring cell behavior including migration in three-dimensional (3-D) tissue cultures. By using a 3-D collagen matrix migration model, we describe an efficient three-stage segmentation algorithm for different cell types. (1) Preprocessing by a rational operator is combined with (2) region growing within a local adaptive window surrounding each cell, and (3) postprocessing with morphological operations. This procedure allows the segmentation of melanoma cells as well as lymphocytes in 3-D collagen matrices independent of the cell shape and position relative to the in-focus plane.