Yan Liping, Chen Yuexia, He Jianxin
Department of Breast Surgery, Maternal and Child Health Hospital of Jiangxi Province, Nanchang, China.
Department of Surgery, the First Affiliated Hospital of Guangxi Medical University, China.
Heliyon. 2024 Sep 28;10(19):e38640. doi: 10.1016/j.heliyon.2024.e38640. eCollection 2024 Oct 15.
A non-invasive predictive model was developed using radiomic features to forecast CXCL9 expression level in breast cancer patients.
CXCL9 expression data and MRI images of breast cancer patients were obtained from The Cancer Genome Atlas (TCGA) and The Cancer Imaging Archive (TCIA) databases, respectively. Local tissue samples from 20 breast cancer patients were collected to measure CXCL9 expression levels. Radiomic features were extracted from MRI images using 3DSlicer, and the minimum Redundancy Maximum Relevance and Recursive Feature Elimination (mRMR_RFE) method was employed to select the most pertinent radiomic features associated with CXCL9 expression levels. Support vector machine (SVM) and Logistic Regression (LR) models were utilized to construct the predictive model, and the area under the receiver operating characteristic curve (AUC) was calculated for performance evaluation.
CXCL9 was found to be upregulated in breast cancer patients and linked to breast cancer prognosis. Nine radiomic features were ultimately selected using the mRMR_RFE method, and SVM and LR models were trained and validated. The SVM model achieved AUC values of 0.748 and 0.711 in the training and validation sets, respectively. The LR model obtained AUC values of 0.771 and 0.724 in the training and validation sets, respectively.
The utilization of MRI radiomic features for predicting CXCL9 expression level provides a novel non-invasive approach for breast cancer Prognostic research.
利用放射组学特征开发一种非侵入性预测模型，以预测乳腺癌患者的CXCL9表达水平。
分别从癌症基因组图谱（TCGA）和癌症影像存档（TCIA）数据库中获取乳腺癌患者的CXCL9表达数据和MRI图像。收集20例乳腺癌患者的局部组织样本以测量CXCL9表达水平。使用3DSlicer从MRI图像中提取放射组学特征，并采用最小冗余最大相关和递归特征消除（mRMR_RFE）方法选择与CXCL9表达水平相关的最相关放射组学特征。利用支持向量机（SVM）和逻辑回归（LR）模型构建预测模型，并计算受试者工作特征曲线下面积（AUC）进行性能评估。
发现CXCL9在乳腺癌患者中上调，并与乳腺癌预后相关。最终使用mRMR_RFE方法选择了9个放射组学特征，并对SVM和LR模型进行了训练和验证。SVM模型在训练集和验证集中的AUC值分别为0.748和0.711。LR模型在训练集和验证集中的AUC值分别为0.771和0.724。
利用MRI放射组学特征预测CXCL9表达水平为乳腺癌预后研究提供了一种新的非侵入性方法。
