摘要
目的 探讨肿块与非肿块型乳腺癌病变采用不同b值行DWI检查,不同操作者间的重复性和一致性.方法 回顾性分析东南大学附属中大医院2010年12月至2017年10月经乳腺动态对比增强MRI（DCE-MRI）检查发现为乳腺非囊性病灶,行乳腺DWI检查,且经外科手术病理证实为乳腺癌的131例女性患者（133个病灶）.肿块型病变94例（96个病灶）,非肿块型病变37例（37个病灶）.所有患者术前均行乳腺DWI检查,80个病灶测量采用4个b值（0、400、800、1000 s/mm2）,其中肿块型病灶有51个,非肿块型病灶29个;53个病灶测量采用5个b值（0、400、800、1200、1600 s/mm2）,其中肿块型病灶45个,非肿块型病灶8个.由2名同等资历的放射科医师采用双盲法阅片,分别测量ADC值.采用配对t检验比较2名医师测量参数的差异,采用组内相关系数（ICC）和Cronbachα系数评价2名医师测量的重复性,采用Bland-Altman一致性分析方法评价2名医师间的一致性.结果 不同b值（0、400、800、1200、1600 s/mm2）下乳腺非肿块型和肿块型病变的ADC值,2名医师间测量结果的差异均无统计学意义（P均〉0.05）.2名医师测得的不同b值下乳腺非肿块型病变的ADC值ICC分别为0.72、0.78、0.87、0.76、0.74,重复性为好或非常好;Cronbachα值分别为0.83、0.88、0.93、0.86、0.85,信度高.2名医师测得的不同b值下乳腺肿块型病变的ADC值ICC分别为0.88、0.95、0.97、0.93、0.91,重复性非常好;Cronbachα值分别为0.94、0.97、0.99、0.96、0.95,信度高.Bland-Altman图像分析结果显示不同操作者在不同b值下测得的乳腺非肿块型、肿块型病变的ADC值一致性均较好.乳腺非肿块型病变在不同b值下,分别有97.3％（36/37）、94.6％（35/37）、93.1％（27/29）、8/8、8/8的点在95％一致性界限内;乳腺肿块型病变在不同b值下,分别有97.9％（94/96）、94.8％（91/96）、92.2％（47/51）、95.6％（43/45）、93.3％（42/45）的点在95％一致性界限内.结论 在不同b值条件下测量乳腺ADC值,不同操作者间对不同病变类型乳腺癌病灶均具有较好的重复性和一致性.
Objective To study the repeatability and consistency among different operators while different b-value DW1 was used to detect mass and non-mass breast lesions. Methods Between December 2010 and October 2017, we collected 131 female patients （one hundred and thirty-three lesions） from Zhongda Hospital Southeast University, they were diagnosed as non-cystic lesion of the mammary gland by contrast enhancement MRI （DCE-MRI）,were examined with DWI and confirmed by surgical pathology. There were 94 cases of mass lesion （96 lesions） and 37 cases of non-mass lesion （37 lesions）. All patients were examined with DWI before surgery,80 patients were examined with b-value of 0, 400, 800 and 1 000 s/mm2, there were 51 tumor lesions and 29 non-tumor lesions;53 patients were examined with b-value of 0, 400, 800 and 1 200,1 600 s/mm2,there were 45 tumor lesions and 8 non-tumor lesions. Two radiologists with the same qualifications used double-blind reading to measure ADC values. Paired t test was used to compare the differences in measurement parameters between the two physicians. The repeatability of data between the two physicians was analyzed by intraclass correlation coefficient （ICC） and Cronbach α coefficients. The consistency of data between the two physicians was evaluated by Bland-Altman analysis. Results In different b values （0,400,800,1 200,1 600 s/mm2）, the ADC value of non-mass type breast cancer and mass type breast cancer was no statistically significance between the two operators （P〉0.05）.The ICC of the ADC values of the non-mass breast lesion under the different b values measured by 2 physicians were 0.72, 0.78, 0.87, 0.76, 0.74, and the repeatability was good or very good; The Cronbach α coefficients are 0.83, 0.88, 0.93, 0.86, 0.85, and the reliability is high. The ICC of the ADC values of the mass breast lesion under the different b values measured by 2 physicians were 0.88,0.95,0.97,0.93,0.91, and the repeatability was very good. the Cronbach α coefficients were 0.94,0.97,0.99,0.96,0.95, and the reliability was high. The results of the Bland-Altman image analysis showed that the ADC values of non-mass and mass lesion detected by different operators in different b values have good consistency. In different b values, there were 97.3% （36/37）, 94.6% （35/37）, 93.1% （27/29）,8/8 and 8/8 points within the 95% consistency limit of non-mass breast lesions; there were 97.9% （94/96）, 94.8% （91/96）, 92.2% （47/51）, 95.6% （43/45）, 93.3% （42/45） within the 95% eonsistency limit of mass breast lesions. Conclusion The ADC values were measured under different b values, and different operators had favarable repeatability and eonsisteney for different lesions of different types of breast eaneer.
作者
王梦茹
刘万花
王瑞
叶媛媛
张悦
Wang Mengru;Liu Wanhua;Wang Rui;Ye Yuanyuan;Zhang Yue(Department of Radiology,Zhongda Hospital,Southeast University,Nanjing 210009,China)
出处
《中华放射学杂志》
CAS
CSCD
北大核心
2018年第8期598-602,共5页
Chinese Journal of Radiology
基金
国家重点基础研究发展计划（2014CB744504）
关键词
乳腺肿瘤
磁共振成像
重复性
一致性
Breast neoplasms
Magnetic resonanee imaging
Reproducibility
Consistency
