Study Of Correlation Between Multiparametric MRI Radiomics Model And Axillary Lymph Node Metastasis In Breast Cancer

Aims Breast cancer is the most prevalent cancer worldwide,and the status of axillary lymph nodes in breast cancer patients affects their treatment decisions and prognosis.It has been a challenge that accurately and non-invasively assessing whether and how many axillary lymph nodes have metastasized.Although the shift from traditional axillary lymph node dissection（ALND）to sentinel lymph node biopsy（SLNB）has prevented some patients from unnecessary axillary surgery,SLNB still has false-negative results or postoperative complications.In recent years,radiomics has rapidly evolved to assist in diagnosing disease and making treatment decisions through a wealth of information not visible to the naked eye in imaging.Magnetic Resonance Imaging（MRI）of the breast is a common imaging test for breast cancer,which the imaging histological features may correlate with axillary lymph node status.The purpose of this study is to identify the imaging histological features associated with axillary lymph node metastasis in breast cancer and to use these features to develop a predictive model to help clinicians provide an effective method to accurately assess the status of axillary lymph nodes before surgery.Materials and Methods 1.We collected the data from 190 breast cancer patients who underwent MRI scan + diffusion + enhancement from January 2017 to December 2021 on the department of breast surgery in the second hospital of Jilin University.2.We generated 3D image in ITK-SNAP software according to the ROI from T2WI-SPAIR、DWI（b=1000）、DCE sequences of breast MRI.3.We imported the images with ROI to A.K and IPMS v2.4.0 software for analyzing,and finally found the characteristic features,which were associated with the lymph node metastasis.4.Apply radiomics pictures to develop a predictive model to distinguish no axillary lymph node metastasis,1-2,3 or more axillary lymph node metastasis.Then obtain ROC curve,AUC value,sensitivity,specificity,and other indicators to evaluate the prediction model’s diagnostic efficacy.Results 1.We developed a predictive model using 9 omics features related to lymph node metastasis to decide whether there is axillary lymph node metastasis using T2WI-SPAIR,DWI（b=1000）and DCE-MRI third-stage sequence.Its AUC value was 0.877（95%CI,0.804-0.942）,the sensitivity and specificity were 0.848 and 0.857 respectively.2.We constructed a predictive model using 5 omics features to differentiate between 1-2 and 3 or more axillary lymph node metastasis according to T2WI-SPAIR,DWI（b=1000）and DCE-MRI third-stage sequence.Its AUC value was 0.841（95%CI,0.822-0.912）,the sensitivity and specificity were 0.820 and 0.805 respectively.Conclusions 1.9 radiomics features are associated with axillary lymph node metastasis and 5radiomics features are concerned with the numbers of the metastatic axillary lymph node based on screening from breast MRI images by radiomics methods.2.The reliable predictive model constructed by applying breast MRIT2WI-SPAIR sequence and DWI（b=1000）sequence combined with DCE-MRI phase III enhancement sequence can be used for clinical non-invasive methods to assess the status of axillary lymph node metastasis and the numbers of metastases,which can optimize current treatment strategy for breast cancer and provide some theoretical basis for clinicians’ preoperative treatment decisions.