改進三維醫學影像重建的混合式內插法

為了讓病人在照電腦斷層掃瞄或類似影像時所吸收的輻射劑量降到最低，我們一般會限制連續影像取樣之二維橫切面間隔在4mm至8mm之間或者更大。當我們使用一序列的影像來重建三維的模型時，若使用影像內插技術來縮小這個距離，則重建後的立體影像的品質將會更好。線性內插（linear interpolation）和形狀內插（shape-based interpolation）是兩種使用最多的內插方法。線性內插可以很容易的實作，並且處理的時間很短，但是內插出來的影像品質不是很理想。另一方面，形狀內插能夠產生較好的影像品質，但是必須花比較多的時間。這篇論文提出了三種結合線性內插與形狀內插兩者優點的混合方法。我們的實驗針對灰階影像進行五個主要的執行步驟：取閥值，距離轉換，線性內插，二元轉換，內插距離的增大。研究結果顯示，處理後的影像品質與處理內插時所花的時間都有顯著的改進，使得醫生在診療時，能夠更容易、更快速與更精確。

關鍵字

形狀內插；線性內插；距離轉換；三維影像重建

並列摘要

To minimize patients' radiation exposure while taking CT images, one normally limits the distance of two consecutive images between 4 mm ~ 8 mm or even larger. Using image interpolation techniques to reduce the gap between image slices can enhance the quality of the reconstructed 3D models. Linear interpolation and shape-based interpolation are the two most popular approaches. While linear interpolation is easy to implement with a short processing time, the quality of the generated image is relatively poor. On the other hand, shape-based interpolation can produce better quality images with much more time. This paper presents three hybrid interpolation methods that take the advantages of the above two approaches. Five main processes are applied to grey-level medical images in our experiments: thresholding, distance transform, linear interpolation, binary conversion, and enlarge increment. The results show significant improvements on 3D image quality and processing speed that can help doctors make their diagnoses easier, faster, and more precise.

並列關鍵字

Shape-Based Interpolation ； Linear Interpolation ； Distance Transform ； 3D Image Reconstruction

國際替代計量

改進三維醫學影像重建的混合式內插法

全文下載

主題瀏覽