REAL-TIME VLSI CHIP FOR 3-D SURFACE RENDERING FOR CT & M

描述：(改编自申请者摘要)两者的可视化 从连续切片中生成的骨和软组织结构 计算机断层扫描(CT)和磁共振(MR)是一种重要的工具 现在用于改进整形外科和许多其他医疗 程序，包括诊断和外科手术。可视化中的 实时，即任何三维视图的即时呈现 根据用户的需求，需要昂贵的计算机图形工作站。 申请人已经获得了一种新的计算结构的专利， 一种基于曲面法线的面绘制方法 使用多端口内存的FCC(面中心立方体)细分。 这应该允许1000万到1亿的实时表面渲染 每秒体素数。 与另一家小企业视觉信息技术公司的讨论， Inc.(德克萨斯州普莱诺)，这是一家图像处理VLSI芯片设计公司， 表明在系统和芯片方面都有精心规划的研究计划 设计优化可以将这种新的计算结构减少到单个 奇普。然后，这些芯片中的一个或多个可能成为 便宜(不到1000美元)。Apple Macintosh或 类似的计算平台。 拟议的研究将包括以下研究：(1)设计权衡 FCC查找表的体系结构和三维 位片图像存储单元；(2)体素值路由和 I/O；(3)主机的选择；(4)覆盖 将计算出的灰度数据存入刷新存储器；以及(5)市场机会 这将恰当地将对专利处理器的研究与 当前国内和国际上对三维表面的需求 渲染。

DESCRIPTION: (Adapted from Applicant's Abstract) Visualization of both bony and soft-tissue structures generated from serial sections produced by computed tomography (CT) and magnetic resonance (MR) is an important tool now employed to improve orthopedic surgery and many other medical procedures, including both diagnostic and surgical. Visualization in real-time, i.e., instantaneous presentation of any three dimensional view upon demand of the user, requires expensive computer graphics workstations. The applicant has received a patent on a new computing structure for surface rendering based on a novel scheme based on using surface normals in the FCC (Face-Centered-Cubic) tessellation employing multiported memory. This should permit real-time surface rendering at from 10 to 100 million voxels per second. Discussions with another small business, Visual Information Technologies, Inc. (Plano, Texas) which is an image-processing VLSI chip design house, indicates that a carefully planned research program on both system and chip design optimization could reduce this new computing structure to a single chip. One or more of these chips could then become the heart of an inexpensive (less than $1000.) plugin board for the Apple Macintosh or similar computing platform. The proposed research would include studies of: (1) design tradeoffs of the architecture of the FCC lookup tables and the three-dimensional bit-slice image memory units; (2) all aspects of voxel value routing and I/O; (3) the choice of the host computer; (4) methods for overlaying the computed graylevel data into refresh memory; and (5) market opportunities which would properly marry the research on the patented processor with current national and international needs in three-dimensional surface rendering.