Fast Parametric Active Contour to Improve 3D Segmentation of Tubular Structures

Segmentation of three dimensional (3D) image data is an important task in medical imaging. The segmentation of tubular structures like e.g. colons, blood vessels or bronchioli is, thereby, an important subtask. Qualitative segmentation of subpleural alveoli in microscopic videos (time as the third dimension) is the objective of the present study. The proposed approach attempts to improve a given segmentation that was obtained by a two dimensional parametric active contour algorithm with an additional smoothness criterion for the third dimension. Compared to 2D segmentation alone, the developed approach decreased the high-frequency noise and resolved problems due to the ambiguity of image-features.