基于成员体关系的医学数据剥离绘制算法

Opacity-Peeling for Medical Data Based on Volume Membership

针对修改传输函数或者使用特殊的光照效果时不能完全解决直接体绘制医学数据人体组织间的遮挡问题,提出一种基于成员体关系的分层剥离体绘制算法.利用梯度阈值函数判断采样点的类型,通过采样光线获得同种组织的标量值范围;用由移动最小二乘法按照拟合出的不同组织的分界标量值和空间位置来建立每个体素的成员体关系;将成员体关系用于分层剥离体绘制,以解决同类组织的遮挡问题,并采用自适应的光线投射算法提高图像的绘制效果和速度.实验结果表明,文中算法可以根据医学数据来确定相应的成员体关系,使被遮挡部分绘制效果清晰.

A transfer function or local illumination is an accepted method for exploring the region of interest for direct volume rendering of medical data.However,this method cannot clearly classify individual tissue and it also cannot render occluded areas.In this paper,we present a method to classify the volume data by creating its membership volume data,which subsequently allows us to display different data layers.A gradient threshold function is introduced to determine the type of a sample in the volume data,and the voxels are then segmented based on the statistical analysis of the whole volume data.The classified volume data is used to establish the membership volume among different tissues according to the scalar range and their spatial information.Finally,we use layer peeling with the membership volume to explore specific tissues,and the occlusion among the same type of tissues is solved.In addition,an adaptive ray-casting algorithm is employed to improve the quality and speed.The experiments demonstrate that our approach can produce volume membership for different medical data and the specific occluded areas can be rendered.