乳腺良恶性病变的3.0T磁共振扩散加权成像鉴别诊断及b值优化

Value of Diffusion-weighted Imaging with Different b Values in the Diagnosis of Breast Lesions at 3.0T MRI

目的:探讨3.0T扩散加权成像(DWI)在乳腺良恶性病变鉴别诊断中的诊断价值并对b值使用进行优化。方法:34例患者共20个良性和16个恶性乳腺病灶纳入研究,同期20例正常乳腺受检者列为对照组。采用3.0T MR行乳腺DWI成像检查,b值分别取0,750 s/mm2,1000 s/mm2,1250 s/mm2。比较不同b值DWI图像质量,测量病灶的对比噪声比(CNR)以及表观扩散系数(ADC)值,以病理诊断为金标准,计算诊断乳腺癌的敏感度、特异度及准确性,绘制受试者工作特征曲线(ROC)并进行比较。结果:不同b值时DWI图像主观质量评分没有统计学差异(F=3.02,P=0.0516>0.05)。b=750 s/mm2时病灶的CNR明显优于b=1000s/mm2及1250s/mm2时(P值为0.004及0.000),而b=1000s/mm2与1250s/mm2之间没有统计学差异(P=0.800>0.05)。相同b值时,正常乳腺组织的平均ADC值>乳腺良性病变>乳腺恶性病变(P均=0.000)。以各b值恶性病变平均ADC值95%可信区间的上限作为界定乳腺癌ADC的阈值,则b=750 s/mm2、1000 s/mm2、1250s/mm2时,ADC阈值分别为1.33×10-3mm2/s、1.25×10-3mm2/s、1.16×10-3mm2/s。以上述阈值判断乳腺癌的敏感度、特异度及准确性分别为81.25%、87.5%、85.71%,81.25%、77.5%、78.57%及81.25%、82.5%、82.14%。三个b值的ROC曲线下面积无显著性差异(P=0.1925>0.05)。结论:乳腺的3.0T DWI中,以b=750s/mm2时病灶CNR最高,值得推荐应用。应用DWI鉴别乳腺良恶性病变时,需要结合ADC值判断。

Purpose： To evaluate the value of diffusion- weighted imaging with different b values in the diagnosis of breast lesions at 3.0T MRI. Methods： Twenty normal controls and 34 patients with 20 benign breast lesions and 16 malignant ones underwent diffusion-weighted imaging at 3.0T MRI with different b values of 0, 750, 1000 and 1250s/mm2. The other conventional sequences （T2WI and T1WI dynamic contrast- enhanced MRI） were performed at 3.0T or 1.5T MRI. Image quality score, contrast-to-noise ratio （CNR） and apparent diffusion coefficient （ADC） of normal breast, benign lesions and malignant ones with the same b value and different b values were compared and analyzed. Sensitivity, specificity and accuracy for the diagnosis of breast cancer with different b values were calculated respectively, as well as the ROC curves. Results： （1） The difference of the image quality score among the different b values was not statistically significant （F=3.02, p=0.0516〉0.05）. （2） The difference of CNR of the breast lesion among the different b values was statistically significant （F=10.69, p=0.0001〈0.01）. Higher CNR was obtained when b value was 750s/mm2 than those when b value was 1000s/mm2 and 1250s/mm2, and the differences were statistically significant （P values were 0.004 and 0.000）, but the difference of CNR between the b value of 1000s/mm2 and 1250s/ram2 was not statistically significant （P=0.800〉0.05）. （3） Average ADC value of normal breast was higher than benign lesion, while that of benign lesion was higher than that of malignant ones at the same b value. Among them the differences were statistically significant （P values were all 0.000）. （4） The cutoff ADC value was 1.33x10 3mm2,1.25x10 3mm2/s, and 1.16x10 3mm2/s when b value was 0,750s/mm2, 0,1000 and 0,1250s/mm2. Sensitivity, specificity and accuracy for diagnosis of breast cancer was 81.25%, 87.5%, 85.71%~ 81.25%, 77.5%, 78.57% and 81.25%, 82.5%, 82.14% respectively. And the area under ROC curve （AUC） was 0.8781, 0.8672 and 0.8828 respectively. The difference among them was not statistically significant （P〉0.05）. Conclusion： （1） Better image quality and CNR score were obtained when b value was 750s/mm2 than those when b value was 1000s/ram2 and 1250s/mm2 at 3.0T. （2） Diagnosis of breast lesions with diffusion-weighted imaging must be made in combination with ADC value. （3） Accuracy in the detection of breast cancer at 3.0T was high with all b values. However we believe the best b value was 0,750s/mm2 because of its highest image quality and CNR.