基于大规模鼠脑血管显微图像的多分辨面绘制可视化方法  

A Multi Resolution Surface Rendering Visualization Method Based on Large Scale Mouse Brain Vascular Microscopic Images

作　　者：饶骁驰 杨昊 周航 文武[1] 李宇昕 RAO Xiaochi;YANG Hao;ZHOU Hang;WEN Wu;LI Yuxin(School of Computer Science,Chengdu University of Information Technology,Chengdu 610225,China;School of Computer Science and Engineering,Xi'an University of Technology,Xi'an 710048,China)

机构地区：[1]成都信息工程大学计算机学院,四川成都610225 [2]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《软件导刊》2025年第2期155-162,共8页Software Guide

基　　金：国家自然科学基金项目(82102137);四川省科技计划项目(2022YFS0542);大学生创新创业训练计划项目(202310621137,202310621147)。

摘　　要：光学成像技术对鼠脑进行大范围甚至全脑显微成像受现有计算机软硬件技术的限制,在GB级大规模三维血管图像的三维模型建模和可视化处理时存在渲染实时性差、可视化处理时间慢的问题。为此,提出包含血管多分辨度建模、视锥可视化成像、IO高性能优化3个方面的多分辨度可视化数据方法。首先,对成像数据进行血管分割后使用Marching Cubes算法和面片削减算法对血管数据进行多分辨度建模;其次,基于视锥剔除思路建立低成本、高实时性的多分辨可视化成像;最后,利用并行化方法进行IO高性能优化。实验表明,所提算法能在733s内快速加载、处理2.23GB的三维血管数据,帧率和并行效率分别为39FPS与23.71%,既提升了模型的可视化速度和响应性,又克服了大规模血管数据可视化中的计算限制,还能更好地展现血管结构。The optical imaging technology for large-scale or even whole brain microscopic imaging of mouse brain is limited by existing computer software and hardware technology.There are problems with poor real-time rendering and slow visualization processing time when modeling and visualizing GB level large-scale 3D vascular images.To this end,a multi-resolution visualization data method is proposed,which includes three aspects:tube multi-resolution modeling,cone visualization imaging,and IO high-performance optimization.Firstly,the imaging data is segmented into blood vessels,and then the Marching Cubes algorithm and patch reduction algorithm are used to perform multi-resolution modeling on the blood vessel data;Secondly,based on the idea of cone removal,a low-cost and high real-time multi-resolution visualization imaging is established;Finally,parallelization methods are used for high-performance optimization of IO.Experiments have shown that the proposed algorithm can quickly load and process 2.23 GB of 3D vascular data within 733 seconds,with frame rates and parallel efficiencies of 39 FPS and 23.71%,respectively.This not only improves the visualization speed and responsiveness of the model,but also overcomes the computational limitations in large-scale vascular data visualization,and better displays the vascular structure.

关 键 词：VTK 多分辨 可视化 医学图像 高性能优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]