利用多解析度全變差最小化技術減少電腦斷層影像之雜訊

電腦斷層攝影（computed tomography, CT）能產生高解析度的切面解剖影像,如今,已廣泛應用於臨床醫療檢查,成為不可或缺的診斷利器。然而,當管電流（mAs）不足時量子斑駁雜訊將導致影像對比及解析度下降。本研究提出一套多解析度全變差最小化演算法（Multi-resolution total variation minimization, MRTV）,首先以Haar小波轉換（wavelet transform）進行影像多解析度拆解,並對高低頻影像以適當的調整參數與迭代次數進行全域之全變差最小化計算,經由反小波轉換即可重建出雜訊抑低之影像。此外,本研究亦提出一套以擬合法快速找出各研究案例之對應調整參數,以提高本研究於臨床應用之可行性。本實驗之研究案例分別為：含有高斯雜訊之Shepp-Logan數位假體與圓球數位假體,圓柱型水假體、Rando假體、頭部、胸部及腹部之低劑量CT影像。結果顯示本研究所提出之MRTV演算法可有效改善量子雜訊之影響,應用於臨床上應可改善影像品質並提升診斷價值。此外,臨床CT若以較低之輻射劑量掃描,經MRTV降噪後將可得到與正常劑量掃描相同之效果,如此可大幅降低病人輻射劑量。

關鍵字

電腦斷層攝影；量子斑駁；小波轉換；全變差最小化

並列摘要

Computed tomography (CT) produces high-resolution images, and had been widely used in clinical examination. However, quantum mottle will decreased the image contrast and resolution, when the tube current (mAs) is insufficient. In this study, a multi-resolution total variation minimization algorithm (MRTV) was proposed. The CT image was firstly decomposed by 2D haar wavelet into the coefficients at high and low frequency. In these coefficients, small local gradients were then removed by total variation minimization with appropriate tunning parameter. Final, the noise reduced image was obtained by inverse wavelet transform of gradient removed coefficients. Since the tuning parameter λ will strongly influence the denoising result, the λ - ρ relation which constructed by linear fitting of best tuning parameters at different noise levels was also proposed to determine tuning parameter for the denoising of any noisy image.In this study, digital phantoms added with Gaussian noise and real CT images acquired with low tube current were used to evaluate the performance of our algorithm. The results showed that the proposed MRTV algorithm can effectly reduced the image quantum noise.The diagnostic value of CT image should be improve by appling our algorithm in clinical.

並列關鍵字

Quantum mottle ； Computed tomography ； Wavelet Transform ； Total variation minimization

參考文獻

謝金宏, 陳淑敏, 黃武達, 劉育朋, 張榮華 and 楊斐適 2006 三度空間多排偵測器電腦斷層血管攝影在新生兒游離肺病患的應用與評估中華放射線技術學雜誌 30 5

Anand C S and Sahambi J S 2010 Wavelet domain non-linear filtering for MRI denoising Magnetic Resonance Imaging 28 842-61

Defrise M and Clack R 1994 A cone-beam reconstruction algorithm using shift-variant filtering and cone-beam backprojection Medical Imaging, IEEE Transactions on 13 186-95

Donoho D L and Johnstone J M 1994 Ideal spatial adaptation by wavelet shrinkage Biometrika 81 425-55

Flohr T G, Bruder H, Stierstorfer K, Petersilka M, Schmidt B and McCollough C H 2008 Image reconstruction and image quality evaluation for a dual source CT scanner Medical Physics 35 5882-97

國際替代計量

利用多解析度全變差最小化技術減少電腦斷層影像之雜訊

未授權

主題瀏覽