论文部分内容阅读
目的:探讨超声影像组学定量特征对浸润性乳腺癌激素受体表达的预测价值。方法:回顾性分析204例浸润性乳腺癌患者的术前超声图像及术后病理结果。根据雌激素受体(estrogen receptor,ER)、孕激素受体(progesterone receptor,PR)及人表皮生长因子受体2(human epidermal growth factor receptor 2,HER-2)表达,将患者分为两组:激素受体阳性组(ER~+、PR~+、HER-2~-),激素受体阴性组(ER~-、PR~-、HER-2~-)。两名具有5年以上临床经验的超声科医师对乳腺癌肿块超声图像进行回顾性特征分析与评估,评估内容包括肿块的形态、边缘、内部回声、后方回声改变及钙化。然后,对同一个肿块利用基于相位信息的动态轮廓模型进行边缘分割。通过t检验,筛选出与激素受体相关性最强的特征参数,通过支持向量机分类器,运用径向基核函数进行分析与研究。为减少偏倚,采用留一法对分类性能进行验证。结果:激素受体阳性组与阴性组在形态、边缘毛刺成角、内部回声及后方回声改变等二维特征方面存在显著统计学差异(P<0.05)。定量分析选出54个定量特征,对激素受体表达具有较高准确率(准确率67.7%,曲线下面积73.2%)。此外,边缘、内部回声、后方回声及钙化等定量特征在激素受体阳性与阴性组之间均存在显著统计学差异(P<0.05)。结论:超声影像组学定量特征分析降低了传统超声影像的主观性,在预测浸润性乳腺癌激素受体表达方面具有较大优势,其在提高超声影像学特征对乳腺癌精准诊断及生物学行为预测方面的价值仍需进一步研究。
Objective: To investigate the predictive value of quantitative features of ultrasound imaging on the expression of hormone receptor in invasive breast cancer. Methods: A retrospective analysis of 204 cases of invasive breast cancer patients with preoperative ultrasound images and pathological results. The patients were divided into two groups according to the expression of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER-2) : Hormone receptor positive group (ER ~ +, PR ~ +, HER-2 ~ -), hormone receptor negative group (ER ~ -, PR ~ -, HER-2 ~ -). Two ultrasonographers with more than 5 years of clinical experience performed retrospective characterization and assessment of ultrasound images of breast masses, including tumor morphology, margins, internal echo, changes in posterior echo, and calcification. Then, using the dynamic contour model based on phase information for the same lump, the edge segmentation is performed. By t-test, the strongest correlation with hormone receptor was screened, and the support vector machine (SVM) classifier was used to analyze and study the radial basis function. In order to reduce the bias, the retention method is adopted to verify the classification performance. Results: There were significant differences in two-dimensional features such as morphology, edge burr angle, internal echo and posterior echo in the hormone receptor positive group and the negative group (P <0.05). Quantitative analysis of 54 quantitative characteristics selected, the hormone receptor expression with high accuracy (accuracy 67.7%, 73.2% under the curve area). In addition, quantitative features such as margins, internal echo, back echo, and calcification were significantly different between the positive and negative hormone receptor groups (P <0.05). Conclusion: The quantitative analysis of ultrasound imageology reduces the subjectivity of traditional ultrasound images and has great advantages in predicting the expression of hormone receptor in invasive breast cancer. It is of great value in enhancing the accurate diagnosis and biological behavior of breast cancer The value of prediction still needs further study.