腦中風是國內十大死亡主因之一。且有逐年增加的趨勢。本研究利用各種不同的影像分析技術，針對磁振照影術獲取的頭部腦組織影像進行處理，以半自動化技術精確的描繪出腦組織損傷區域。希望能夠及早提供資訊，便於進行臨床診斷與治療。研究過程使用不同的演算法則〔Sobel演算法，Marr-Hildreth演算法，Canny演算法〕進行腦組織損傷區域描繪，其中以Canny演算法結果與真實腦組織損傷區域之吻合度最高，也能精確的描繪出腦組織損傷區域和鄰近正常腦組織或是顱骨的界線，是研究時可靠的檢測模式。Canny演算法與醫師人爲手動的描繪出腦組織損傷區域比較，結果顯示出Canny演算法的可靠度和精確度比人爲方式來的迅速又準確，是研究時方便又可靠的檢測模式。

Diffusion-weighted MR Imaging is the most important clinical tools for diagnosis of acute brain infraction. This research takes advantage of all sorts of different image analysis technology that direct to brain infraction images of Magnetic Resonance proceed dispose, and automatization technology precise brain tissue damage area to replace artificial manual trouble and not precise. In the study, three different edge detection algorithms (namely Sobel algorithm, Marr-Hildreth algorithm and Canny algorithm) have been applied as preprocessing for brain infraction outlining detection. Among all these algorithms, Canny algorithm result and real brain tissue damage region inosculate highest, and it cans precise trace brain infraction area and normal brain tissue. Canny algorithm with doctors' artificial manual depiction brain infarction area, Canny algorithm's dependable degree and definition than artificial style to rapidly right again. Therefore, the Canny algorithm eventually achieve the best outline contour.