头颈部病变MR扩散成像的初步研究

Application of diffusion weighted echo-planar MR imaging in head and neck lesions

目的探讨 MR 扩散加权成像(DWI)在头颈部病变中的诊断价值。方法回顾性分析57例头颈部病变共85个病灶,其中恶性肿瘤22个(22例),良性肿瘤13个(13例),囊性或液性病灶13例共17个病灶,包括囊肿12个(8例),肿瘤坏死4个(4例),脓肿1个(1例);淋巴结病灶33个。所有患者均经临床追踪或手术病理证实。分析各病灶在扩散敏感因子(b 值)为0、500和1000 s·mm^(-2)时 DWI 特点及其表观扩散系数(ADC)值。结果良、恶性肿瘤在 DWI 表现有所不同,13个良性肿瘤与脊髓(或脑干)的信号比值随 b 值增加迅速下降;22个恶性肿瘤与脊髓(或脑干)信号比值随 b 值增加变化不明显。恶性肿瘤的平均 ADC 值[(0.78±0.24)×10^(-3)mm^2·s^(-1)]低于良性肿瘤[(1.48±0.20)×10^(-3)mm^2·s^(-1)],差异有统计学意义(t=8.9,P<0.01);良、恶性肿瘤的 ADC值受试者特异性曲线(ROC)下面积 Az 值为0.971±0.030,当 ADC 值取1.13×10^(-3)mm^2·s^(-1)时,判断良、恶性肿瘤的敏感度为100%,特异度为90.5%,准确度90.5%。鳞状细胞癌的 ADC 值与分化程度相关,分化好者其 ADC 值高。囊性或液性病变(12个)、肿瘤坏死(4个)和脓肿(1个)在 DWI 上的信号强度与其组织病理成分相关,具有一定的信号特征和 ADC 值。淋巴结转移瘤的平均 ADC 值[(0.81±0.11)×10^(-3)mm^2·s^(-1)]高于良性淋巴结病变[(0.69±0.04)×10^(-3)mm^2·s^(-1)],差异有统计学意义(t=3.48,P<0.01),但是两者之间有较大的重叠,且在 DWI 上表现相仿。结论 DWI 和ADC 值在头颈部病变的鉴别诊断中具有重要价值。

Objective to investigate the value of diffusion-weighted echo-planar MR imaging in the diagnosis of head and neck lesions. Methods Fifty-seven patients with 85 head and neck lesions were enrolled in the study, including 22 patients with 22 malignant tumors, 13 patients with 13 benign tumors, 13 patients with 17 cystic and liquefactive lesions（ including 8 patients with 12 cystic lesions ,4 patients with 4 tumor necrosis, 1 patients with 1 abcess） and 33 lymph nodes. The lesions were all confirmed by operation and clinical follow up. Echo-planar diffusion-weighted imaging （DWI） was performed with different b values （0,500, and 1,000 s·mm^-2）, and the apparent diffusion coefficients （ADCs） were measured. Results Malignant and benign tumors had different characteristics in DWI with different b values. With the increase of b value, the signal intensity of tumor/spinal cord ratio decreased quickly in DWI in benign tumors, while the signal intensity of tumor/spinal cord ratio remained similar in DWI in malignant tumors. The mean ADC value of malignant tumors [ （0. 78 ± 0. 24）×10^-3 mm^2·s^-1 ] was significantly lower than that of benign tumors [ （ 1.48 ± 0. 20）×10^-3 mm^2·s^-1 ] （ t = 8.9, P 〈 0. 01 ） ; The areas under the receiver operating characteristic （ROC） curve in discriminating malignant from benign tumors was 0. 971 ± 0. 030. When an ADC value smaller than 1.13 × 10^-3 mm^2·s^-1 was used for predicting malignancy, the sensitivity, specificity, and accuracy were 90.5%, 100%, and 90.5%, respectively. The ADCs of squamous cell carcinoma （SCC） correlated with its histopathologic differentiation, well differentiated SCCs had higher ADC values. The signal intensities of cystic and liquefactive lesions, tumor necrosis and abscess were related with their histopathologic constituents, and they had special signal characteristics and ADC values. The mean ADC value of metastatic lymph nodes [ （0. 81 ±0. 11 ） ×10^-3mm^2 ·s^-1 ] was significantly higher than that of benign lymphadenopathy [（0.69±0.04）×10^-3mm2·s^-1] （t =3.48, P 〈0.01）, but they had overlaps and similar appearances. Conclusion DWI and the measurement of ADCs are useful in characterizing the head and neck lesions.