奈米物體立體資料之即時導覽系統

Abstract

體積顯影技術(Volume Rendering)在目前醫學影像上已經是非常廣泛且常用的一種方法，本篇論文是熟悉這項技術並且加以實做。其中在裡面添加類似遊戲控制和細節層次(level of detail, LOD)的功能，可使使用者更加容易上手操作，並加上一些視覺化的效果，讓使用者可以去調整想觀看的區域且隨著視野拉近可以看到不同細緻度的貼圖效果，而達到觀察研究的目的。這個程式主要是應用在醫學影像上，所以多半是用來觀察不同器官血管結構，通過這些設計最終是想讓使用者可以像是任意遊走在器官中，觀察整個影像重建出來後的效果。 隨著電腦掃描技術越來越好，影像解析度越來越高，圖片的檔案相對的就越來越大，在成像繪圖時，GPU的運算量相對也會越來越大，但當遊走在影像間時，其實有很多東西是再觀察者背面而看不到的，在這次的設計中，有加入檔案切割的部分，使觀察時只載入可觀察到的檔案影像，減少繪圖時GPU的計算量，希望使整個影像觀察上能更佳的流暢。

Volume Rendering is one of a widely used method in medical imaging. The main idea of this thesis is to be familiar with Volume Rendering and bring it into practice as a program with extra functionalities. Details include, adding gaming control function, and level of detail (LOD) function to build a user friendly control environment. In addition, adding some visual effects options allow user to change the region of interest and to observe different LOD while zooming in and out. All of above help scientists analyze volume medical images. The program is mainly design for observing details of medical image, such as blood vessels structure from different organs. The goal of this thesis is to observe the result and quality of reconstructed volume image of an organ by allowing user to navigate in it. The size and resolution of a medical image increase due to the improvement of scanning technology. It increases the computational effort of GPU when rendering the image. In fact, there is a large portion of volume behind the sight of observer which cannot be seen and can be ignore at the time while navigating in volume image. In this thesis, we cut a big volume into small cubes and save them as separated files. Than only load those cubes (files) that are in the sight of observer to reduce computing quantity for GPU in order to make the navigating process smoother.