体素内不相干运动联合扩散峰度成像模型对乳腺良恶性病灶的鉴别诊断价值

The value of IVIM-DKI model in differentiating benign from malignant breast lesions

目的：探讨体素内不相干运动（IVIM）技术联合扩散峰度成像（DKI）对乳腺良恶性病变的鉴别诊断价值。方法：137例女性患者（共153个病灶）行双侧乳腺多b值DWI检查（b=0~2000s/mm2）;分别使用IVIM、DKI模型获得病灶的真性扩散系数（D）、灌注相关扩散系数（D＊）、灌注分数（f）以及平均扩散峰度系数（MK）、平均扩散系数（MD）和ADC值。分析这些参数在乳腺良恶性病变中的变化规律,采用受试者工作特性曲线（ROC）评估各参数的诊断效能。结果：良恶性病灶的D、f、MK、ADC和MD值的中位数差异有统计学意义（P值分别为0.000、0.020、0.000、0.000和0.000）,良恶性病灶D＊值的差异无统计学意义（P=0.480）。D值与和MK、ADC和MD值在鉴别乳腺良恶性病变中的可靠性相当,两两比较差异无统计学意义（P值均大于0.1）。当MK值及D、ADC和MD值的阈值分别取0.8073及0.9536×10-3、1.1436×10-3和1.5657×10-3 mm2/s时,鉴别良恶性病灶的敏感度和特异度依次分别为（95.7%,84.2%）及（95.7%,81.6%）、（96.5%,84.2%）和（93.9%,84.2%）。ROC曲线分析得出D值的诊断效能最大（AUC=0.91）。联合D值和MK值的AUC达0.92。结论：采用IVIM和DKI模型获得的相关参数有助于乳腺良恶性病灶的鉴别,以IVIM模型中的真性扩散系数的诊断敏感性和特异性较高,联合真性扩散系数和扩散峰度系数的诊断效能最高。

Objective：To study the diagnostic value of IVIM-DKI model MRI in differentiating benign from malignant breast lesions. Methods： 137 female patients （153 lesions in total） underwent multi b value （b=0-2000s/mm^2） DWI scanning. Tissue diffusivity （D） ,pseudo-diffusivity （D＊） ,perfusion fraction （f） and mean diffusion kurtosis （MK） ,mean diffusivity （MD） and ADC were calculated using IVIM and DKI model. The difference of these parameters in benign and malignant breast lesions was analyzed statistically. The diagnostic performance of these parameters was evaluated by ROC analysis. Results：There were statistically significant differences between malignant and benign lesions in the diffusion parameters of median D, f, MK, ADC and MD value （P= 0. 000,0. 020,0. 000,0. 000 and 0. 000 respectively）. There was no statistically significant difference in D＊ value （P=0. 480）. The reliability of D, MK, ADC and MD value for differential diagnosis of be- nign and malignant breast lesions was similar, there was no significant difference between any two of them （P~ 0. 05）. When the optimal threshold of MK,D,ADC and MD value was 0. 8073,and 0. 9536）〈 10 a, 1. 1436 X 10 a and 1. 5657 X 10 atom2/s, the corresponding diagnostic sensitivity and specificity were 95.7 %, 84.2 %;95.7%, 81.6%;96.5 %, 84.2% and 93.9% ,84.2% ,respectively. ROC analysis showed that the AUC of D value was 0.91 which was the highest one. By combining D with MK value for diagnosis, AUC was increased to 0.92. Conclusion： The parameters obtained from IVIM- DKI model are helpful to differential diagnosis of benign from malignant breast lesions. The D value obtained from IVIM model may have relatively high sensitivity and specificity,D Combined with MK value can obtain the best diagnostic performance.