MRI影像组学在早期浸润性乳腺癌腋窝淋巴结转移诊断的价值

Value of MRI in the Diagnosis of Axillary Lymph Node Metastasis of Early Invasive Breast Cancer

目的分析MRI影像组学诊断早期浸润性乳腺癌腋窝淋巴结转移的价值。方法分析2018年12月至2020年12月在本院诊治的96例早期浸润性乳腺癌的临床资料,患者均接受MRI动态增强扫描。分析96例患者的腋窝淋巴结增强影像,由两名医师分别评估图像并开展组内及组间一致性评估;96例患者按病理结果分腋窝淋巴结转移阴性(阴性组,n=49)、腋窝淋巴结转移阳性(阳性组,n=55),分层抽样法将两组淋巴结分为训练集及验证集,特征降维算法选择有统计学意义的MRI影像组学参数,ROC曲线分析MRI影像组学参数对早期浸润性乳腺癌的预测效能。结果提取6大类共401个影像组学参数,降维后仅直方图参数-均匀度、形态学参数-表容比、纹理特征参数-全角度集群突出方差、灰度共生矩阵参数-全角度相关.性、游程矩阵参数-长行程优势有统计学意义(P<005),直方图参数-均匀度预测早期浸润性乳腺癌的曲线下面积.(AUC)最高.,并显著高于形态学参数-表容比(Z=3304、P=0.001)、纹.理特征参数-全角度集群突出方差(Z=2495.,P=0012).、灰度共生矩阵参数-全角度相关性(Z=5.212、P<0.001)、游.程矩阵参数-长行程优势(Z=4623、P<0001),以08018为Cut-off,其.预测早期.浸润性乳腺癌的敏感度、特异度分别为9455%、9796%。结论MRI影像组学可作为临床评估早期浸润性乳腺癌腋窝淋巴结转移状态的可靠方式,尤其是直方图参数-均匀度,具良好的敏感度及特异度,值得临床重视。

Objective The aim of this study was to analyze the value of MRI in the diagnosis of axillary lymph node metastasis of early invasive breast cancer. Methods The clinical data of 96 patients with early invasive breast cancer diagnosed and treated in the hospital were analyzed between December 2018 and December 2020. All patients enrolled were subjected to dynamic contrast-enhanced MRI scan. The axillary lymph node enhanced images of the 96 patients were analyzed and evaluated by two physicians. According to pathological results, the patients were divided into axillary lymph node metastasis negative group(negative group, n=49) and axillary lymph node metastasis positive group(positive group, n=55). Using stratified sampling method, the two groups of lymph nodes were divided into a training set and a validation set. The feature dimension reduction algorithm was used to select statistically significant MRI parameters, and the ROC curve was used to analyze the predictive efficiency of MRI parameters for early invasive breast cancer. Results A total of 401 imaging parameters in 6 categories were collected. After dimension reduction, only histogram parameter-uniformity, morphological parameter-surface volumetric ratio, texture feature parameter-all-angle cluster highlighting variance, gray-level co-occurrence matrix parameter-all-angle correlation, and run-length matrix parameter-long stroke advantage showed statistically significant differences(P<0.05). The area under the curve(AUC) of histogram parameter-uniformity was the largest for predicting early invasive breast cancer, which was significantly larger than that of morphological parameter-surface volumetric ratio(Z=3.304, P=0.001), texture feature parameter-all-angle cluster highlighting variance(Z=2.495, P=0.012), gray-level co-occurrence matrix parameter-all-angle correlation(Z=5.212, P<0.001), and runlength matrix parameter-long-stroke advantage(Z=4.623, P<0.001). With 0.801 8 as the cut-off value, the sensitivity and specificity in predicting early invasive breast cancer were 94.55% and 97.96%, respectively. Conclusion MRI can be used as a reliable way to assess the status of axillary lymph node metastasis of early invasive breast cancer in clinical practice, especially the histogram parameteruniformity, which has good sensitivity and specificity.