Non-invasive prediction of the tumor growth rate using advanced diffusion models in head and neck squamous cell carcinoma patients.
Non-invasive prediction of the tumor growth rate using advanced diffusion models in head and neck squamous cell carcinoma patients.
复制标题
DOI：
10.18632/oncotarget.16851
复制
发表时间：
2017-05-16
期刊：
影响因子：
--
通讯作者：
Shirato H
中科院分区：
文献类型：
--
作者：
Fujima N;Sakashita T;Homma A;Harada T;Shimizu Y;Tha KK;Kudo K;Shirato H
我们评估了55例头颈部鳞状细胞癌(HNSCC)患者的先进弥散加权成像(DWI)模型预测肿瘤生长速度的参数。弥散加权成像采集使用单次激发自旋回波回波平面成像，具有12个b值(0−，2000年)。我们使用单指数、双指数、三指数、拉伸指数和扩散峰度成像模型计算了14个DWI参数。我们从两组不同日期的成像数据中直接测量了肿瘤的生长率。我们根据患者获得MR的日期将患者分为发现组(n=40)和确认组(n=15)。在发现组中，我们进行了单因素和多因素回归分析，建立了使用扩散参数预测肿瘤生长率的多元回归方程。将发现组得到的方程应用于验证组，以验证方程的准确性。对发现组患者进行单因素和多因素回归分析后，利用三指数模型得到的中间扩散系数D2和缓慢扩散系数D3的显著参数，建立了估计的肿瘤生长率方程。发现组估计的肿瘤生长率与直接测量的肿瘤生长率之间的相关系数为0.74。在验证组中，估计的肿瘤生长率与直接测量的肿瘤生长率之间的相关系数(r=0.66)和类内相关系数(0.65)分别为良好。综上所述，先进的DWI模型参数可以作为预测HNSCC患者肿瘤生长速度的指标。
We assessed parameters of advanced diffusion weighted imaging (DWI) models for the prediction of the tumor growth rate in 55 head and neck squamous cell carcinoma (HNSCC) patients. The DWI acquisition used single-shot spin-echo echo-planar imaging with 12 b-values (0−2000). We calculated 14 DWI parameters using mono-exponential, bi-exponential, tri-exponential, stretched exponential and diffusion kurtosis imaging models. We directly measured the tumor growth rate from two sets of different-date imaging data. We divided the patients into a discovery group (n = 40) and validation group (n = 15) based on their MR acquisition dates. In the discovery group, we performed univariate and multivariate regression analyses to establish the multiple regression equation for the prediction of the tumor growth rate using diffusion parameters. The equation obtained with the discovery group was applied to the validation group for the confirmation of the equation's accuracy. After the univariate and multivariate regression analyses in the discovery-group patients, the estimated tumor growth rate equation was established by using the significant parameters of intermediate diffusion coefficient D2 and slow diffusion coefficient D3 obtained by the tri-exponential model. The discovery group's correlation coefficient between the estimated and directly measured tumor growth rates was 0.74. In the validation group, the correlation coefficient (r = 0.66) and intra-class correlation coefficient (0.65) between the estimated and directly measured tumor growth rates were respectively good. In conclusion, advanced DWI model parameters can be a predictor for determining HNSCC patients’ tumor growth rate.