Building a 3D statistical model of the C3-6 inter-vertebral disks using spine MR images; a comparison between PCA & ICA

In this paper, we construct a 3D statistical model of the Inter-vertebral Disks using whole body spine MR images. After building the initial atlas in natural coordinate system, the landmarks extracted from it are propagated to all samples of all data-sets. PCA and ICA are utilized to explore the linear variation of landmarks. The results of PCA and ICA are compared with each other by calculating the compactness capacity, specificity and generalization ability measures. The results show that the model built by PCA is more compact and for a constant number of modes, the reconstruction error is significantly lower for statistical models built by PCA. Also, in sufficient number of modes to make the model as general as possible, the error made by specificity measure is lower for PCA. Therefore, PCA produces a statistical model with substantially better specificity.

کلیدواژه ها:

component ، statistical Atlas Construction ، Statistical Shape Model ، Linear PCA ، ICA ، Inter-vertebral Disks ، Spine MRI