Introduction Specific Aim Volumetric Datasets 3D Bilateral Filtering EM based Classification Classification and Segmentation Collaborators

Images are pre-filtered by Perona-Malik's method, and then segmented/classified by the fast marching method. The critical points are detected by the anisotropic vector diffusion, and classified by peak-searching of the histogram. All methods used for classification, filtering, and segmentation are developed in CCV, The University of Texas at Austin.

2D Result

Original image: slice #00	Classification	Segmentation	Hand-picking result
Original image: slice #10	Classification	Segmentation	Hand-picking result
Original image: slice #20	Classification	Segmentation	Hand-picking result
Original image: slice #30	Classification	Segmentation	Hand-picking result
Original image: slice #40	Classification	Segmentation	Hand-picking result
Original image: slice #50	Classification	Segmentation	Hand-picking result
Original image: slice #60	Classification	Segmentation	Hand-picking result

3D Result