Kuru Timur H, Roethke Matthias C, Stieltjes Bram, Maier-Hein Klaus, Schlemmer Heinz-Peter, Hadaschik Boris A, Fenchel Michael
From the *Department of Urology, University Hospital Heidelberg, Heidelberg, Germany; †Department of Radiology, ‡Quantitative Imaging-based Disease Characterization, and §Medical and Biological Informatics, German Cancer Research Center (DKFZ), Heidelberg, Germany.
J Comput Assist Tomogr. 2014 Jul-Aug;38(4):558-64. doi: 10.1097/RCT.0000000000000088.
To compare 2 previously presented algorithms for extracting parameters from intravoxel incoherent motion (IVIM) studies and investigate them in the context of tissue differentiation.
Magnetic resonance imaging (MRI) was performed in 23 patients without histologically proven prostate carcinoma (PCa) and 27 patients with histologically proven PCa. Two methods were used to determine IVIM parameters (f, D, D*). Receiver operating characteristic analysis was performed for IVIM parameters and apparent diffusion coefficient for discrimination of prostate tissue.
The IVIM parameters showed no significant difference between patients without PCa and normal areas in patients with PCa (r = 0.46-0.99). Results for D were not significantly different for both methods (P = 0.22), whereas f from method 1 was significantly higher than the f from method 2 (P < 0.05). The diffusion parameters D (both methods) and apparent diffusion coefficient could discriminate between tumor and normal areas (receiver operating characteristic analysis, area under the curve, ≥0.90). Additionally, in subgroup analysis, only D was able to discriminate between low- and high-grade PCa.
For tumor detection, IVIM diffusion does not yield a clear added value, but the perfusion-free diffusion constant D may hold potential for improved image-based tumor grading.
比较两种先前提出的从体素内不相干运动(IVIM)研究中提取参数的算法,并在组织分化的背景下对其进行研究。
对23例无组织学证实前列腺癌(PCa)的患者和27例有组织学证实PCa的患者进行磁共振成像(MRI)检查。使用两种方法确定IVIM参数(f、D、D*)。对IVIM参数和表观扩散系数进行受试者操作特征分析,以鉴别前列腺组织。
IVIM参数在无PCa的患者和有PCa患者的正常区域之间无显著差异(r = 0.46 - 0.99)。两种方法得到的D结果无显著差异(P = 0.22),而方法1得到的f显著高于方法2得到的f(P < 0.05)。扩散参数D(两种方法)和表观扩散系数能够区分肿瘤和正常区域(受试者操作特征分析,曲线下面积≥0.90)。此外,在亚组分析中,只有D能够区分低级别和高级别PCa。
对于肿瘤检测,IVIM扩散没有产生明显的附加值,但无灌注扩散常数D可能具有改善基于图像的肿瘤分级的潜力。