Wang He, Krishnan Sunil, Wang Xiaochun, Beddar A Sam, Briere Tina M, Crane Christopher H, Mohan Radhe, Dong Lei
Department of Radiation Physics, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA.
Int J Radiat Oncol Biol Phys. 2008 Sep 1;72(1):201-9. doi: 10.1016/j.ijrobp.2008.04.054.
To investigate a deformable image registration method to improve soft-tissue contrast in four-dimensional (4D) computed tomography (CT) images of the liver.
Ten patients with hepatocellular carcinoma underwent 4D CT scan for radiotherapy treatment planning on a positron emission tomography/CT scanner. Four-dimensional CT images were binned into 10 equispaced phases. The exhale phase served as the reference phase, and images from the other nine phases were coregistered to the reference phase image using an intensity-based, automatic deformable image registration method. Then the coregistered images were combined to create a single, high-quality reconstructed CT image at exhale phase as the new reference for target delineation. The extent of image quality enhancement was quantified relative to the original CT by calculating the signal-to-noise ratio and the contrast-to-noise ratio.
The soft tissue image contrast was noticeably better after deformable image registration than in the original scans. Signal-to-noise ratios inside the liver region of interest increased for all patients by a factor of 3.0 (range, 2.3-3.7). The improvement in image quality was not linearly proportionate to the number of images averaged. Using only 6 phases can achieve at least 85% of the contrast enhancement that can be achieved using all 10 phases. We also found that contrast enhancement was inversely proportional to the original image quality (p = 0.006), and the contrast enhancement is attained with little loss of spatial resolution.
This deformable image registration method is feasible to improve soft-tissue image quality in 4D CT images.
研究一种可变形图像配准方法，以提高肝脏四维（4D）计算机断层扫描（CT）图像中的软组织对比度。
10例肝细胞癌患者在正电子发射断层扫描/CT扫描仪上进行4D CT扫描，用于放射治疗计划。四维CT图像被分为10个等间距的时相。呼气时相作为参考时相，使用基于强度的自动可变形图像配准方法将其他九个时相的图像与参考时相图像进行配准。然后将配准后的图像合并，以创建一个呼气时相的高质量重建CT图像，作为靶区勾画的新参考。通过计算信噪比和对比噪声比，相对于原始CT对图像质量增强程度进行量化。
可变形图像配准后软组织图像对比度明显优于原始扫描。所有患者肝脏感兴趣区内的信噪比提高了3.0倍（范围为2.3 - 3.7）。图像质量的改善与平均图像数量并非线性比例关系。仅使用6个时相可实现使用全部10个时相所能达到的至少85%的对比度增强。我们还发现对比度增强与原始图像质量成反比（p = 0.006），并且在空间分辨率几乎没有损失的情况下实现了对比度增强。
这种可变形图像配准方法对于提高4D CT图像中的软组织图像质量是可行的。
