Objective: To investigate the features of apparent diffusion coefficient (ADC) histogram parameters based on entire tumor volume data in high resolution diffusion weighted imaging of nasopharyngeal carcinoma (NPC) and to evaluate its correlations with cancer stages. Methods: This retrospective study included 154 cases of NPC patients[102 males and 52 females, mean age (48±11) years]who had received readout segmentation of long variable echo trains of MRI scan before radiation therapy. The area of tumor was delineated on each section of axial ADC maps to generate ADC histogram by using Image J. ADC histogram of entire tumor along with the histogram parameters-the tumor voxels, ADC(mean), ADC(25%), ADC(50%), ADC(75%), skewness and kurtosis were obtained by merging all sections with SPSS 22.0 software. Intra-observer repeatability was assessed by using intra-class correlation coefficients (ICC). The patients were subdivided into two groups according to cancer volume: small cancer group (<305 voxels, about 2 cm(3)) and large cancer group (≥2 cm(3)). The correlation between ADC histogram parameters and cancer stages was evaluated with Spearman test. Results: The ICC of measuring ADC histogram parameters of tumor voxels, ADC(mean), ADC(25%), ADC(50%), ADC(75%), skewness, kurtosis was 0.938, 0.861, 0.885, 0.838, 0.836, 0.358 and 0.456, respectively. The tumor voxels was positively correlated with T staging (r=0.368, P<0.05). There were significant differences in tumor voxels among patients with different T stages (K=22.306, P<0.05). There were significant differences in the ADC(mean), ADC(25%), ADC(50%) among patients with different T stages in the small cancer group(K=8.409, 8.187, 8.699, all P<0.05), and the up-mentioned three indices were positively correlated with T staging (r=0.221, 0.209, 0.235, all P<0.05). Skewness and kurtosis differed significantly between the groups with different cancer volume(t=-2.987, Z=-3.770, both P<0.05). Conclusion: The tumor volume, tissue uniformity of NPC are important factors affecting ADC and cancer stages, parameters of ADC histogram (ADC(mean), ADC(25%), ADC(50%)) increases with T staging in NPC smaller than 2 cm(3).
目的： 探讨鼻咽癌3.0T MRI高分辨率扩散加权成像表观扩散系数(ADC)整体直方图参数特征及与肿瘤分期的相关性。 方法： 回顾性分析154例新发鼻咽癌患者的MRI资料，其中男102例，女52例，年龄15～80岁，平均(48±11)岁。所有患者治疗前行高分辨率扩散加权成像分段读出平面回波成像序列(RESOLVE)检查。利用Image J软件在ADC图上逐层分割肿瘤区域并生成直方图，最后由SPSS 22.0软件生成整体直方图，1个月后由同一研究者重复上述步骤。利用SPSS软件计算两次测量的组内相关系数(ICC)；生成鼻咽癌的ADC直方图参数(体素数、ADC(mean)、ADC(25%)、ADC(50%)、ADC(75%)、偏度和峰度)，比较不同肿瘤分期直方图参数的差异。按体素数差异，分为小病灶组(<305体素，约2.0 cm(3))、大病灶组(≥2.0 cm(3))。直方图参数与分期的相关性采用Spearman相关分析。 结果： 鼻咽癌ADC整体直方图两次测量体素数、ADC(mean)、ADC(25%)、ADC(50%)、ADC(75%)、偏度、峰度的ICC值分别为0.938、0.861、0.885、0.838、0.836、0.358、0.456，观察者内测量的一致性好(偏度、峰度除外)。不同T分期的鼻咽癌体素数差异有统计学意义(K＝22.306，P<0.05)，与T分期正相关(r＝0.368，P<0.05)。小病灶组ADC(mean)、ADC(25%)、ADC(50%)在T1、T2、T3、T4分期差异均有统计学意义(K＝8.409、8.187、8.699，均P<0.05)，与T分期正相关(r＝0.221、0.209、0.235，均P<0.05)；大病灶组不同T分期ADC值、偏度及峰度值差异均无统计学意义；大、小病灶组间偏度、峰度差异有统计学意义(t＝－2.987，Z＝－3.770，均P<0.05)。 结论： 鼻咽癌肿瘤体积、组织的均匀度是影响ADC值及肿瘤分期的重要因素；小体积鼻咽癌ADC(mean)、ADC(25%)、ADC(50%)与T分期正相关，大体积鼻咽癌各直方图参数在不同肿瘤分期均无统计学意义。.
Keywords: Apparent diffusion coefficient; Histogram; Nasopharyngeal carcinoma; TNM system.