表面容積素化及骨骼肌肉手術的應用

Surface Voxelization with Application on Musculoskeletal Surgeries

林家誠(Chia-Cheng Lin)

指導教授 ： 蔡明達

中原大學/工學院/資訊工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300139

摘要

摘要
現存的容積表示法是使用容積素來建構一個無厚度表面的模型，以此方式並不能正確地操作表面和實體之間的解剖結構改變。本論文的研究目的主要是描述容積的表示式，其中在每一個容積素上使用面碼旗標和距離階值來表示表面的拓撲和幾何結構，並且使用物體旗標來識別實體。在本論文中也提出了一系列的演算法來容積化實體和表面，並且檢查容積表面與固體之間的封閉情形及在固體上施以不同的操作。這個方法可以提供顱顏面和整形外科部門所有外科手術療法的主要功能。
我們提出了下列的演算法來達成本論文之目的
1. 每一個容積素經過正規化步驟後可得到三個軸向的距離階值，其值代表著容積素中心沿著該軸向到同質面的距離。經過正規化過程後，同質面上的樣點只要由該容積素值就可決定，而不用經由相鄰的容積素決定。因此解剖結構可以藉由周圍容積素距離階值的變化來改變位置。
2. 開發3D解剖結構的切割演算法來改變手術刀之切割平面上之容積素的距離階值，以產生新的邊界。
3. 開發高效率的3D種子繁衍演算法來辨識分離的解剖結構。經由此步驟，以便可以對解剖結構做移動、移除和碰撞偵測。
4. 利用3D種子繁衍演算法將解剖結構上的容積素暫存到記憶體中並且清除原解剖結構中的記憶體值，然後再將記憶體中的容積素寫到新的位置上來模擬解剖結構的移動。
5. 使用光跡追蹤演算法來計算解剖結構之邊界容積素的移動資訊，以判斷是否和其它結構產生碰撞情形。
6. 在兩個解剖結構上分別指定融合平面以生成骨頭容積素。先辨識新生成的骨頭容積素，然後再辨識分離的兩個解剖結構而成為一個連續的解剖結構。

關鍵字

容積視覺化 ； 三維影像為基礎的手術模擬 ； 醫學應用

並列摘要

Abstract
Existing volumetric representations model a non-thickness surface using voxels, and thus cannot accurately manipulate topology changes in the interactions between surfaces and solids. This study describes a volume representation in which face flags and distance-levels on each voxel represent surface topology and geometry, and an object flag identifies solids. Also presented herein is a set of algorithms that voxelize solids and surfaces, check the closures of the interactions among the volumetric surfaces and solids and manipulate various operations on the solids. The method can provide the main functions of all surgical modalities used in craniofacial and orthopedic departments.

We implement the following algorithms to achieve this purpose.
1. Normalizing a voxel with three distance-levels. Each distance-level represents the distance from the voxel center to the isosurface along an axis. After the normalization, a sample point on the isosurface can be determined by only one voxel, not by a pair of neighboring voxels; an anatomic structure can be repositioned regarding its surrounding voxels.
2. Using a section algorithm of 3D anatomic structure, then change the distance-levels of voxels in cut plane region of surgical knife to create the new boundary.
3. Using a high efficient 3D seed and flood algorithm to recognize separate anatomic structures. As the result, we can reposition, remove and detect collision of the structures.
4. Extending the 3D seed and flood algorithm to clear the voxels of a structure, then rewrite it to a new position for simulating the reposition of the structure.
5. Using a ray traversal algorithm to compute the traversal information of boundary voxels of an anatomic structure, then determine the collision condition of the structure.
6. Generating bone voxels between specified fusing surfaces of two structures to generate; recognizing the bone voxels, re-recognize one of the two structures to obtain a continuous separate structure.

並列關鍵字

3D image based surgical simulation ； volume visualization ； medical applications

參考文獻

[Chen 85] Lih-Shyang Chen et al., “Surface Shading in th Cuberille

Distributed and Interactive Three-Dimensional Medical Image

System”, Computerized Medical Imaging & Graphics, Vol.18,

[Feng 98] Feng, L. and Soon, S.H., 1998, “An Effective 3D Seed Fill Algorithm”, Computer & Graphics, Vol.22, No.5, pp.641-644.

[Fujimoto 1986] A. Fujimoto et al., “ARTS: Accelerated Ray Tracing System”, IEEE CG&A, April 1986, Vol. 6, No. 4, pp. 16-24.