Morphometry of 4D Eigen Spectral Images for Characterizing Spatio-temporal Properties of Intratumoral Enhancement Patterns in Dynamic Contrast-enhanced Breast MRI

摘要

This study was designed to characterize the spatio-temporal properties of intratumoral enhancement patterns on 4D eigen spectral images for MR-based breast tumor diagnosis. MR-time-series were registered for movement compensation and tumor regions were segmented semi-automatically from our proposed perfusion index map. Eigenvalues were obtained for temporal enhancement curves for each voxel within tumor by using singular value decomposition (SVD), generating eigenvalue maps. The spatial variations of intratumoral eigenvalues were captured by 3D moment invariants. The influence on mophometric features of eigenvalue maps in tumor differentiation is validated by classification performance using linear discriminant analysis and leave-one-out cross validation.