Garrett John, Ge Yongshuai, Li Ke, Chen Guang-Hong
Department of Medical Physics, University of Wisconsin-Madison, Wisconsin 53705.
Department of Medical Physics, University of Wisconsin-Madison, Wisconsin 53705 and Department of Radiology, University of Wisconsin-Madison, Wisconsin 53792.
Med Phys. 2014 Dec;41(12):120701. doi: 10.1118/1.4901313.
In x-ray absorption mammography, it has been found that the anatomical background noise can be characterized by a power law dependence on the spatial frequency, NPSa(f) ≈ αf(-β). In this letter, the authors present the first experimental results of the corresponding exponents, β, for differential phase contrast (βDPC) and dark field contrast (βDF) mammography.
A grating-based x-ray multicontrast imaging acquisition benchtop system was used to simultaneously acquire mammograms with three different contrast mechanisms from 15 cadaver breasts under the same x-ray data acquisition conditions. The cadaver breasts were imaged in the coronal plane. The authors' experimental implementation of the well documented method [Burgess, Jacobson, and Judy, Med. Phys. 28, 419-437 (2001)] to extract the exponent β was first validated using anonymized clinical mammograms. Experiments were then used to determine β for the three types of mammograms for each cadaver breast acquired with our multicontrast imaging system: absorption contrast mammogram (βAbs.), differential phase contrast mammogram (βDPC), and dark-field contrast mammogram (βDF).
The measured β values, acquired in the coronal plane with the benchtop multicontrast imaging system are βAbs. = 3.61 ± 0.49, βDPC = 2.54 ± 0.75, and βDF = 1.44 ± 0.49 for absorption, differential phase, and dark field mammogram, respectively.
The β values for differential phase contrast and dark field mammography are significantly lower than the measured value of β for the corresponding absorption contrast mammograms. The greatly reduced β value of the anatomical background noise in differential phase contrast and dark field mammograms may suggest potentially improved diagnostic performance for certain types of breast cancer imaging tasks.
在X射线吸收乳腺摄影中,已发现解剖学背景噪声可通过对空间频率的幂律依赖性来表征,即NPSa(f)≈αf^(-β)。在本文中,作者展示了差分相衬(βDPC)和暗场相衬(βDF)乳腺摄影相应指数β的首个实验结果。
使用基于光栅的X射线多对比度成像采集台式系统,在相同的X射线数据采集条件下,从15具尸体乳房同时获取具有三种不同对比度机制的乳腺图像。尸体乳房在冠状面成像。作者对文献记载完善的提取指数β的方法[Burgess, Jacobson, and Judy, Med. Phys. 28, 419 - 437 (2001)]的实验实施,首先使用匿名临床乳腺图像进行了验证。然后通过实验确定了用我们的多对比度成像系统为每具尸体乳房采集的三种类型乳腺图像的β值:吸收对比度乳腺图像(βAbs.)、差分相衬乳腺图像(βDPC)和暗场对比度乳腺图像(βDF)。
使用台式多对比度成像系统在冠状面获取的测量β值,吸收、差分相衬和暗场乳腺图像的β值分别为βAbs. = 3.61 ± 0.49、βDPC = 2.54 ± 0.75和βDF = 1.44 ± 0.49。
差分相衬和暗场乳腺摄影的β值显著低于相应吸收对比度乳腺图像的β测量值。差分相衬和暗场乳腺图像中解剖学背景噪声的β值大幅降低,可能表明对于某些类型的乳腺癌成像任务,诊断性能有潜在改善。