Department of Radiology, Hôpital Tenon, Université Pierre et Marie Curie, 4 rue de la Chine, 75020 Paris, France.
Radiology. 2011 Mar;258(3):793-803. doi: 10.1148/radiol.10100751. Epub 2010 Dec 30.
To retrospectively determine the value of adding perfusion-weighted (PW) and diffusion-weighted (DW) sequences to a conventional magnetic resonance (MR) imaging protocol to differentiate benign from malignant tumors.
The institutional ethics committee approved this retrospective study and waived the requirement to obtain informed consent. MR images in 87 women (age range, 25-87 years) who underwent imaging before surgery for complex adnexal masses-excluding endometriomas and cystic teratomas-were analyzed. Conventional morphologic, perfusion, and diffusion MR criteria of malignancy were recorded. Three independent observers reviewed images in four steps: conventional MR images alone, conventional MR images and PW images combined, conventional MR images and DW images combined, and conventional, PW, and DW MR images combined. Receiver operating characteristic curve analysis was performed to compare the results of the readings. A recursive partitioning model was built to establish a multivariate decision tree.
There was almost perfect agreement for lesion characterization regardless of the reader experiment or step considered (κ = 0.811-0.929). Area under the receiver operating characteristic curve values were higher for conventional and DW images combined, conventional and PW images combined, and conventional, DW, and PW images combined compared with conventional MR images alone (P < .05). For all readers, the accuracy of conventional, PW, and DW imaging combined was higher than that of conventional MR imaging alone for benign masses (P < .01) but not for malignant masses (P = .24). The addition of both PW and DW images led to a correct change in the diagnosis in 19% (11 of 57 patients), 23% (13 of 57 patients), and 24% (14 of 57 patients) of cases for readers 1, 2, and 3, respectively, with no incorrect changes. Conventional, PW, and DW MR imaging criteria were combined to generate a decision tree giving an accuracy of 95%.
The addition of PW and DW sequences to a conventional MR imaging protocol improved the diagnostic accuracy in the characterization of complex adnexal masses.
回顾性分析在常规磁共振成像(MR)方案中添加灌注加权(PW)和弥散加权(DW)序列对区分良、恶性肿瘤的价值。
本研究经机构伦理委员会批准,回顾性分析 87 例经手术治疗的复杂附件肿块(不包括子宫内膜异位症和囊性畸胎瘤)患者的术前 MR 图像,患者年龄 25~87 岁,均为女性,且无需知情同意。记录常规形态学、灌注和弥散 MR 恶性肿瘤标准。3 名独立观察者分 4 个步骤对图像进行分析:仅行常规 MR 检查,常规 MR 联合 PW 检查,常规 MR 联合 DW 检查,常规、PW 和 DW 联合检查。采用受试者工作特征曲线分析比较不同检查结果,建立多元决策树。
无论观察者试验或考虑的步骤如何,病变特征的一致性均近乎完美(κ=0.811~0.929)。与仅行常规 MR 检查相比,常规 MR 联合 DW 图像、常规 MR 联合 PW 图像以及常规、PW 和 DW 联合检查的曲线下面积更大(P<0.05)。对于所有观察者,与常规 MR 检查相比,常规、PW 和 DW 联合检查对良性肿块的诊断准确率更高(P<0.01),但对恶性肿块的诊断准确率差异无统计学意义(P=0.24)。1、2、3 号观察者分别在 19%(11/57 例)、23%(13/57 例)和 24%(14/57 例)的病例中,通过添加 PW 和 DW 图像可正确改变诊断,且无错误改变。常规、PW 和 DW 磁共振成像标准联合生成的决策树准确率为 95%。
在常规 MR 成像方案中添加 PW 和 DW 序列可提高复杂附件肿块的诊断准确率。