Department of Radiology, Boston University Medical Center, Boston, MA, USA.
Med Phys. 2012 Oct;39(10):6028-34. doi: 10.1118/1.4752086.
To further improve the image quality, in particularly, to suppress the boundary artifacts, in the extended scan field-of-view (SFOV) reconstruction.
To combat projection truncation artifacts and to restore truncated objects outside the SFOV, an algorithm has previously been proposed based on fitting a partial water cylinder at the site of the truncation. Previous studies have shown this algorithm can simultaneously eliminate the truncation artifacts inside the SFOV and preserve the total amount of attenuation, owing to its emphasis on consistency conditions of the total attenuation in the parallel sampling geometry. Unfortunately, the water cylinder fitting parameters of this 2D algorithm are inclined to high noise fluctuation in the projection samples from image to image, causing anatomy boundaries artifacts, especially during helical scans with higher pitch (≥1.0). To suppress the boundary artifacts and further improve the image quality, the authors propose to use a roughness penalty function, based on the Huber regularization function, to reinforce the z-dimensional boundary consistency. Extensive phantom and clinical tests have been conducted to test the accuracy and robustness of the enhanced algorithm.
Significant reduction in the boundary artifacts is observed in both phantom and clinical cases with the enhanced algorithm. The proposed algorithm also reduces the percent difference error between the horizontal and vertical diameters to well below 1%. It is also noticeable that the algorithm has improved CT number uniformity outside the SFOV compared to the original algorithm.
The proposed algorithm is capable of suppressing boundary artifacts and improving the CT number uniformity outside the SFOV.
进一步提高图像质量，特别是抑制扩展扫描视场（SFOV）重建中的边界伪影。
为了消除投影截断伪影并恢复 SFOV 外的截断物体，先前已经提出了一种基于在截断处拟合部分水圆柱的算法。先前的研究表明，由于该算法强调平行采样几何中总衰减的一致性条件，因此它可以同时消除 SFOV 内的截断伪影并保留总衰减量。不幸的是，该 2D 算法的水圆柱拟合参数在图像之间的投影样本中容易受到高噪声波动的影响，导致解剖边界伪影，特别是在较高螺距（≥1.0）的螺旋扫描期间。为了抑制边界伪影并进一步提高图像质量，作者提出使用基于 Huber 正则化函数的粗糙度惩罚函数来加强 z 维边界一致性。已经进行了广泛的体模和临床测试，以测试增强算法的准确性和稳健性。
在体模和临床病例中，使用增强算法观察到边界伪影的显著减少。与原始算法相比，该算法还降低了水平和垂直直径之间的百分比差异误差至远低于 1%。此外，该算法还提高了 SFOV 外的 CT 数均匀性。
所提出的算法能够抑制边界伪影并提高 SFOV 外的 CT 数均匀性。
