透過您的圖書館登入
IP:3.133.142.193

摘要


利用碎波轉換方法分析具有阿茲海默症狀及正常電腦斷層影像組成份;以此法分析治療中的電腦斷層影像,以便比較治療前和治療中診斷到具有壞死腫瘤區域的影像組成份,做爲臨床判斷及治療方針參考。採用Daubechies(db3)的碎波對具有阿茲海默症狀及正常的電腦斷層影像及分析治療中的電腦斷層的影像的組成份;以碎波分解(wavelet analysis)的方式,以此方法分解首先要有一個原始無窮數列對影像做解壓縮的動作,此數列裡的每個元素的值,可以是影像中每個圖點的灰階值。有了這樣的一個原始數列之後,要將原始數列分別進行低頻分析濾波以及高頻分析濾波這兩個步驟。再以Matlab用二項法(dyadic)對碎波做多重解析,把函數或算子拆解成不同尺度的組合。分析法機本上是要把影像用平移(shifting)及縮放(scaling)對一張圓形做碎波分解的參術探討,在取得參數系數後就可以對組成份做量化的探討。在比較具有阿茲海默症狀及正常的電腦斷層影像的組成份時,發現正常和異常的影像在分解後的係數參數上有顯著的差異;而分析治療中的電腦斷層的影像,於比較治療前和治療中被診斷到具有壞死腫瘤區域的影像組成份時,也發現治療前和治療中被診斷到具有壞死腫瘤區域的影像在分解後的係數參數上,且在特定的係數有顯著的差異。如果以傅立葉級數來處理影像,此級數只處理了2π週期函數:2π週期函數可以被整數頻率的正弦餘弦波所合成。對於週期比2π大的週期函數及對於非週期性的函數就要就要用到傅立葉變換。所以先前以傅立葉變換和傳立葉級數處理圓形時,可以說是從時問域(time domain)轉換到頻率域(frepuency domain)的運算。但是,傅立葉級數所沒有的方便性及特性都可以用碎波(Wavelet)做“平滑化”的哈耳基底的(Haar Basis)處理,這些特性讓我很輕易的把一張影像做抽絲剝繭的處理,把影像李面的組成份用不同的係數參數表達,這是以碎波理論應用到影像分析的重要結論。

並列摘要


The image could be decomposed by wavelet transformation. Mathematically, the process of wavelet transformation is to extract the image approximation and detail coefficients, and plot the position versus scaling for some interesting area to know the composition of the area. Calculating wavelet coefficients at every possible scale is a fair amount of work, and it generates an awful lot of data. We choose scales and positions based on powers of two-so called dyadic scales and positions and we adopted such an analysis method called discrete wavelet transformation to perform level one and level two decomposition by using Matlab for much more efficient and just as accurate analysis. Performing image with one and two step decomposition by using Daubechies wavelet function obtained a set of approximation and detail coefficients. Extract coefficients at level one and level 2 from wavelet decompostion structure and then reconstruct image by attaching a colormap from the wavelet decomposition structure to show a colorful image. The plot of position versus scales for some interesting area reveal to the characteristic of composition in the area. The characteristics of position versus scales plot exhibit two different composition area and the different plot curves pattern could be used to distinguish two unknown lesions successfully.

延伸閱讀