3D shape-dependent thinning method for trabecular bone characterization
The six space directions.
A trabecular bone sample (a), its surface skeleton (b), and its curve skeleton (c).
Global chart of the hybrid skeletonization algorithm.
The two computer-generated test vectors and results of the two shape classification algorithms. Plate elements are shown in light gray while the rod elements of the original volume are in dark gray.
Results of the MESPTA thinning algorithm on the two hybrid shaped test vectors. Plates are well described while rod elements are not sufficiently eroded.
Results of the Betti numbers curve thinning algorithm on the two hybrid shaped test vectors. Rods are well described while plates are not preserved.
Results of the hybrid skeleton algorithm on the two hybrid shaped test vectors. The surface skeleton is in light gray while the curve skeleton is in dark gray.
Two extracts from OA and OP samples to illustrate the micro architectural differences in the two trabecular bones. The solid phase (bone) is in black while the pore phase is in white.
An extract of 2003 voxels of a trabecular OP bone sample (a), its surface (b), curve (c), and hybrid (d) skeletons.
An extract of a trabecular bone sample (a), its corresponding hybrid skeleton (b), and classified volume (c). Plate trabeculae are shown in light gray while rod trabeculae are in dark gray.
The Percentage of incorrectly classified rod, plate, and overall voxels (PICrV, PICpV, and oPICV, respectively) as a measure for the shape classification efficiency of the MAC and SCA algorithms applied on test vectors 1 and 2.
Values of β0, β2, and N3, measured on both the original object and its hybrid skeleton for each of the two test vectors.
Results of the comparative study on different skeletonization methods using two sets of trabecular bone samples. Mean ± standard deviation and the Student |t| values for each feature estimated from the curve, surface and hybrid skeletons.
Results of the comparative study on the two sets of trabecular bone samples. Mean ± standard deviation and the Student |t| values for each feature estimated from the SCA-based classified volumes.
Article metrics loading...
Full text loading...