本研究計畫的主要目地是在發展生物及醫學影像的基礎電腦輔助幾何分析方法。現代生物及醫學造影科技，像是顯微鏡、電腦斷層、及核磁共振，可以產生極佳高解析度影像圖給醫師做更正確的早期病理診斷。但是，高解析度影像圖也意謂著圖檔資料的資料量也會跟著大量增加。傳統的醫學影像處理方法主要是以局部的畫素分析方法為主，所以所需的處理時間是相對於資料量的大小而線性的增加。相較之下，只有少數的研究是在找出整個器官的幾何結構的全面分析及瞭解。在此研究計畫中，我們計畫要開發此一未知的領域，我們特別會將重心放在研發血管、腦、及大腸等器官的幾何結構的分析方法。一般血管形狀像樹狀結構可以中軸線及半徑來當數學模型分析，而腦及大腸可以參數式的繪圖來當數學模型分析。如此一來，高維的資料像是二維或三維的血管影像資料可以被簡化為一維曲線，三維的腦及大腸的體積資料可以被簡化為二維的參數式的繪圖。這些簡化後的數學模型可以讓我們發展更有效率的自動幾何分析方法來萃取有用的生醫資訊以協助病情診斷。我們希望能研發更好的數學方法來協助電腦科學家及生物醫學工程師開發更好的電腦輔助診斷軟體來協助醫師診斷病情。最終是希望能增進全民的健康福祉。This research project is aimed at developing fundamental computer-aided geometric analysis methods for biomedical image processing. Modern biomedical imaging technologies such as microscopy, computed tomography (CT), and magnetic resonance imaging (MRI) have provided medical doctors with unprecedentedly high-resolution images to improve the diagnostic accuracy of diseases at a much earlier stage. However, high resolution images also mean that the size of data sets increases dramatically. Traditional medical image processing methods mostly rely on texture analysis to derive localized characteristics. Therefore, their processing time increases linearly with the size of data sets. Less research was done on developing global geometric information extraction for high resolution data sets to provide a higher level of understanding of the overall shape and structure of studied organs. In this project, we intend to fill this knowledge gap and are particularly interested in developing geometric methods to help study the global structure of blood vessels, brain folding, and the colon. The blood vessels can be modeled as tree structures with centerlines and varying diameters, while the brain and the colon can be modeled by parametric surfaces. In such a way, higher dimensional data such as 2D or 3D blood vessel images can be simplified to 1D curves and 3D brain and colon volumetric data sets can be simplified by 2D parameterized surfaces. These simplified models allow more efficient geometric analysis methods to extract useful information to evaluate the condition of a patient’s health. Our goal is to develop better mathematical methods that computer scientists and biomedical engineers can use to develop better computer-aided diagnosis systems to help medical doctors diagnose diseases. Ultimately we hope to improve the general health care quality of the nation. 研究期間：10008 ~ 10107
