影像组学对不同X线表现类型乳腺病灶良恶性的鉴别诊断效能

Differential Diagnostic Performance of Mammography Radiomics in the Differential Diagnosis of Benign and Malignant Breast Lesions for Various X-ray Phenotype

目的探讨影像组学特征分析方法对不同X线表现类型乳腺病灶良恶性的诊断效能。资料与方法回顾性分析行乳腺X线摄片检查并获得病理诊断结果的816例女性患者资料。运用手动分割的方法分割病灶感兴趣区,并进行影像组学特征预处理和提取,用五折交叉验证及逻辑回归方法进行良恶性病灶分类器训练和测试。使用受试者工作特征曲线下面积(AUC)评估乳腺病灶良恶性的诊断效能。结果影像组学特征分类器对乳腺肿块型病灶的鉴别诊断能力最强,其次是钙化,对非对称和结构扭曲的鉴别诊断效能较低(AUC值依次为0.82±0.02、0.75±0.07、0.61±0.05、0.58±0.10,P<0.05);在放射科医师阅片基础上联合运用影像组学方法可提高结构扭曲类型乳腺病灶的鉴别诊断效能(AUC值由0.78±0.08上升至0.82±0.08,P<0.05)。结论影像组学特征分析对于鉴别X线表现类型肿块型病灶的良恶性具有较好的诊断价值。

Purpose To evaluate the differential diagnostic performance of mammography radiomics for various X-ray phenotype of breast lesions.Materials and Methods Retrospectively analyzed the clinical data of 816 female patients who underwent mammography examination before obtained the pathological diagnosis results.Manual segmentation was used to segment the region of interest of the lesion,the image preprocessed and feature extraction were conducted concurrently.Five-fold-cross-validation and Logistic regression were used to train and test the classifier for distinguishing benign and malignant lesions.The area under the receiver operating characteristic curve(AUC)was used to evaluate the diagnostic efficacy.Results The radiomic classifier had the strongest ability in the differential diagnosis of mass lesions,followed by calcification,and had low efficiency in the differential diagnosis of asymmetry and structural distortion(with the AUC values of 0.82±0.02,0.75±0.07,0.61±0.05,0.58±0.10,respectively,P<0.05).On the basis of radiologists'reading films,combined use of the radiomic methods can improve the differential diagnosis efficiency for structural distortion(AUC increased from 0.78±0.08 to 0.82±0.08,P<0.05).Conclusion Mammography radiomics show good differential diagnostic performance of the mass type lesions of X-ray manifestations.