由心臟斷層掃描自動診斷冠狀動脈斑塊之技術

Abstract

近年來，由於多切面電腦斷層冠狀動脈攝影(MDCT)的快速發展，電腦斷層影像變的更精確、解析度更高，而被廣泛使用。然而，一位病人的電腦斷層掃描影像動輒上百張，使得醫師需耗費大量的時間與精神才能診斷出冠狀動脈是否健康，因此本論文希望能達成由電腦自動診斷冠狀動脈斑塊，以減少醫師診斷時所耗費的時間成本與人力成本。 本篇論文主要提出了兩種自動化診斷冠狀動脈斑塊之方法，第一種方法是利用傅立葉描述子與亮度特徵來診斷冠狀動脈斑塊，而第二種方法則針對血管亮度與半徑進行迴歸分析，來達到自動診斷冠狀動脈斑塊。其中，我們提出的第二種方法，有較高的Sensitivity與P.P.V.，且相較於另外11位研究者提出的方法也有不錯的成績。

In recent years, due to the rapid development and expanded use of MDCT(multi-detector computed tomography), it becomes much easier than before to obtain more accurate and high-resolution image data of the heart. In general, a patient's MDCT data consist of hundreds of slices and it requires a lot of time and effort to generate diagnoses from such data. Our goal is to propose and analyze methods for diagnosing coronary plaques automatically. In this thesis, we propose two automated method for the diagnosis of coronary artery plaques. To identify potential coronary artery plaques, the first method uses Fourier descriptors and intensity features, and the second method uses regression analysis on the intensities and radii of vascular cross-sections. Our second method ranks in second place when compared with the results of 11 other algorithms listed on the data provider's website.