使用倒傳遞類神經網路於自動化雙邊濾波腦部磁振影像

在取得磁振影像的過程中，經常伴隨著隨機雜訊的干擾，進而影響影像的品質，而本研究所使用的雙邊濾波器便為一種廣泛被使用之去雜訊技術，其能有效地去除磁振影像中的隨機雜訊，提升影像品質。但應用此濾波器需要由使用者嘗試輸入各種不同的濾波參數值組合，做參數的最佳化調整，以獲得最佳濾波結果。然而，這種測試過程是十分花費時間的，並且不易得知所重建的影像是否為最佳品質。因此，為了解決此問題，本論文使用磁振影像的特徵資料，結合倒傳遞類神經網路訓練出一個可預測參數之模型，使用此模型得到最佳化雙邊濾波參數值的設定，進而自動化移除腦部磁振影像中的雜訊。本研究應用灰階共生矩陣以及離散小波轉換等方法擷取影像特徵，並使用T檢定方法選取能有效分辨資料差異性的特徵。我們使用一系列T1權重的模擬磁振影像來測試此一自動化去雜訊系統。實驗結果顯示本研究所提的方法能有效地預測參數，並且自動化去除磁振影像中的雜訊。此方法所建立的預測模型，在濾波參數的預測上有良好的準確性，最後獲得的重建影像亦有相當不錯的品質。

關鍵字

磁振影像；雙邊濾波器；倒傳遞類神經網路；灰階共生矩陣；小波轉換； T檢定

並列摘要

The bilateral filter has been widely used in many image processing applications. It is an effective filtering algorithm that can remove the random noise in magnetic resonance (MR) images. However, the bilateral filter requires the end-user to try different combinations of parameter values in order to obtain the optimal filtering results. This testing process is very time-consuming and difficult to know whether the reconstructed images have the optimal quality or not. To solve this problem, this thesis proposes using the MR image features in combined with a back propagation neural network to establish a predictable parameter model. The goal is to use this model to optimize parameters settings and to automate the denoising procedure. We adopt the gray level co-occurrence matrix and the discrete wavelet transform method for image features extraction. The T-test method is then used to select the features that can effectively distinguish characteristics differences in image data. We have used a wide variety of simulated T1-weighted MR images to evaluate the proposed automatic denoising system. The experimental results indicated that the proposed method effectively predicted the bilateral filtering parameters and automatically removed the noise in MR images. In summary, this new method creates a prediction model with high predictive accuracy and produces reconstructed images with good quality both qualitatively and quantitatively.

並列關鍵字

magnetic resonance image (MRI) ； back-propagation neural network ； bilateral filter ； gray level co-occurrence matrix (GLCM) ； discrete wavelet transform ； T-test

參考文獻

[23] 賴哲儇, "高光譜影像立方體於特徵空間之三維紋理計算," 碩士, 國立中央大學, 2009.

[3] A. Macovski, "Noise in MRI," Magnetic Resonance in Medicine, vol. 36, pp. 494-497, 1996.

[4] H. Gudbjartsson and S. Patz, "The Rician distribution of noisy MRI data," Magnetic Resonance in Medicine, vol. 34, pp. 910-914, 1995.

[5] R. D. Nowak, "Wavelet-based Rician noise removal for magnetic resonance imaging," Image Processing, IEEE Transactions on, vol. 8, pp. 1408-1419, 1999.

[6] L. He and I. R. Greenshields, "A nonlocal maximum likelihood estimation method for Rician noise reduction in MR images," Medical Imaging, IEEE Transactions on, vol. 28, pp. 165-172, 2009.

國際替代計量

使用倒傳遞類神經網路於自動化雙邊濾波腦部磁振影像

全文下載

主題瀏覽