Tissue classification based on 3D local intensity structures forvolume rendering

摘要

This paper describes a novel approach to tissue classificationnusing three-dimensional (3D) derivative features in the volume renderingnpipeline. In conventional tissue classification for a scalar volume,ntissues of interest are characterized by an opacity transfer functionndefined as a one-dimensional (1D) function of the original volumenintensity. To overcome the limitations inherent in conventional 1Dnopacity functions, we propose a tissue classification method thatnemploys a multidimensional opacity function, which is a function of then3D derivative features calculated from a scalar volume as well as thenvolume intensity. Tissues of interest are characterized by explicitlyndefined classification rules based on 3D filter responses highlightingnlocal structures, such as edge, sheet, line, and blob, which typicallyncorrespond to tissue boundaries, cortices, vessels, and nodules,nrespectively, in medical volume data. The 3D local structure filters arenformulated using the gradient vector and Hessian matrix of the volumenintensity function combined with isotropic Gaussian blurring. Thesenfilter responses and the original intensity define a multidimensionalnfeature space in which multichannel tissue classification strategies arendesigned. The usefulness of the proposed method is demonstrated byncomparisons with conventional single-channel classification using bothnsynthesized data and clinical data acquired with CT (computedntomography) and MRI (magnetic resonance imaging) scanners. Thenimprovement in image quality obtained using multichannel classificationnis confirmed by evaluating the contrast and contrast-to-noise ratio innthe resultant volume-rendered images with variable opacity values