以最小生成樹為基礎的醫學影像分割法

Abstract

影像分割的問題，在醫學影像的應用上，是常被討論的課題，並且在做疾病分析及生物研究上有極大的價值。藉由在影像中以每個格子點當作是節點，相鄰的格子點當作邊，本論文提出一種以最小生成樹為基礎的影像分割方法。為了提升其在醫學影像上的適用性，我們整合了區域之間的整體相異度與局部相異度計算，並實際應用到不同的性質的醫學影像，其結果在正確性和速度上都產生令人滿意的結果。

The problem of image segmentation is an important task in medical imaging applications , and it is of great worth in disease analysis and biological research. In this thesis , we present a method of image segmentation using minimum spanning trees by taking each pixels in the image as nodes and each adjacency pixels as edges. In order to improve its feasibility in medical images , we integrate the computation of global variation and local variation. When we apply it to medical images of different properties in practice , the result in correction and speed is satisfying.