基于MRI联合特征鉴别鼻腔鼻窦恶性黑色素瘤与鳞癌  

Differentiation of sinonasal malignant melanoma and squamous cell carcinoma based on MRI combined features

作　　者：郑璇 赵会军 王新艳[1] 张涵娟 彭如臣[2] 鲜军舫[1] ZHENG Xuan;ZHAO Hui-jun;WANG Xin-yan;ZHANG Han-juan;PENG Ru-chen;XIAN Jun-fang(Department of Radiology,Beijing Tongren Hospital Affiliated to Capital Medical University,Beijing 100000,China;Department of Radiology,Beijing Luhe Hospital Affiliated to Capital Medical University,Beijing 101100,China)

机构地区：[1]首都医科大学附属北京同仁医院放射科,北京100000 [2]首都医科大学附属北京潞河医院放射科,北京101100

出　　处：《中国临床医学影像杂志》2023年第12期851-855,共5页Journal of China Clinic Medical Imaging

基　　金：北京市医院管理中心“登峰”计划专项经费资助(编号:DFL20190203)。

摘　　要：目的:鼻腔鼻窦恶性黑色素瘤易误诊为鳞癌,但二者治疗方法不同,故本研究探讨MRI特征在鉴别恶性黑色素瘤与鳞癌中的能力。方法:回顾性分析122例经病理证实为鼻腔鼻窦恶性黑色素瘤及鳞癌的患者的临床及MRI资料,比较两组肿瘤的MRI征象,采用多因素Logistic回归分析获取具有鉴别诊断价值的MRI征象,并采用受试者工作特征(ROC)曲线分析其诊断恶性黑色素瘤的灵敏度、特异度和准确率,建立Logistic回归模型构建联合参数,评价其诊断效能。结果:鼻腔鼻窦恶性黑色素瘤与鳞癌在性别、肿块部位、形态、T_(1)WI呈高信号、T_(2)WI呈低信号、T_(1)WI分隔征(T_(1)-SP)、坏死、眼眶受累及面颊软组织受累方面有统计学差异(均P<0.05)。多因素Logistic回归分析结果显示肿块形态、T_(1)WI呈高信号、T_(1)-SP、坏死是恶性黑色素瘤的独立影响因素,其中T_(1)WI呈高信号准确率最高,为83.6%。ROC曲线分析结果显示包括T_(1)WI呈高信号、T_(1)-SP、坏死及肿块形态在内的MRI特征联合预测恶性黑色素瘤的诊断效能最高,灵敏度、特异度和准确率分别为85.5%、90.6%和86.9%,高于任一单一指标(P<0.001)。结论:MRI特征有助于鉴别鼻腔鼻窦恶性黑色素瘤与鳞癌,Logistic回归模型有助于提高恶性黑色素瘤的诊断率,为临床诊断提供依据。Objective:Sinonasal malignant melanoma is easily misdiagnosed as squamous cell carcinoma,but the treatment methods are different.Therefore,this study explores MRI features of sinonasal malignant melanoma as compared with squamous cell carcinoma.Methods:Retrospective analysis was conducted for 122 patients with complete clinical and MRI data,confirmed by surgery and pathology,including 69 cases with sinonasal malignant melanoma(SMM)and 53 patients with squamous cell carcinoma(SCC).We compare the differences of MRI features between SMM and SCC.The logistic regression analysis was performed to identify the most discriminating MRI features.And the sensitivity,specificity,and accuracy of diagnosing malignant melanoma were analyzed by receiver operating characteristic(ROC)curve.The model of multiple parameters was established by logistic analysis,and the diagnostic efficacy was evaluated.Results:The gender,tumor location,shape,hyperintense on T_(1)WI,hypointense on T_(2)WI,septate pattern on T_(1)WI(T_(1)-SP),necrosis,orbital involvement and cheek soft tissue involvement were significantly different between sinonasal malignant melanoma and squamous cell carcinoma(P<0.05).Logistic regression analysis showed mass shape,hyperintense on T_(1)WI,T_(1)-SP and necrosis were the independent influencing factor of sinonasal malignant melanoma,hyperintense on T_(1)WI has the highest accuracy,which is 83.6%.The sensitivity,specificity and accuracy of the combination of mass shape,hyperintense on T_(1)WI,T_(1)-SP and necrosis were 85.5%,90.6%and 86.9%,respectively.The diagnostic performance of combination features was higher than that of each single sign(P<0.001).Conclusion:MRI features are helpful to distinguish sinonasal malignant melanoma from squamous cell carcinoma.The logistic regression model helps to improve the diagnostic rate of sinonasal malignant melanoma,providing evidence for clinical diagnosis.

关 键 词：鼻肿瘤 黑色素瘤 磁共振成像 

分 类 号：R739.62[医药卫生—肿瘤] R445.2[医药卫生—临床医学]