Background: Imageology uses high-throughput and automatic computing methods to transform medical image data into quantitative data with feature space, and then makes accurate quantitative analysis, extracts features and builds models, which can intuitively observe the overall features of lesions and the surrounding tissues, and provide rich invisible information. At present, the research on the imaging features of dynamic contrast-enhanced (DCE) and diffusion-weighted imaging (DWI) to predict the molecular typing value has achieved results, but the imaging model based on DWI and DCE-magnetic resonance imaging (MRI) is not enough to predict the molecular subtypes, and the prediction value of the prediction model based on the three-dimensional volume of interest of the lesion to the four molecular subtypes of breast cancer has not been fully studied.
Methods: The clinical data of 202 breast cancer patients at our hospital from October 2020 to November 2021 were collected. All patients were examined with multimodal MRI before surgery. Base on immunohistochemical recombinant Ki-67 protein (Ki-67), estrogen receptor (ER), human epidermal growth factor receptor-2 (HER-2) and progesterone receptor (PR) results, the tumors were divided into four types According to the results of the sentinel lymph node (SLN) biopsies, the patients were divided into SLN (+) and SLN (-) groups. 3-dimensional (3D) Slicer software was used to outline the region of interest (ROI), and AMni-Kinetics software was used for feature extraction. The imaging characteristics were screened using least absolute shrinkage and selection operator (LASSO)-Logistic regression model using R statistical software, and the receiver operating characteristic (ROC) curve was drawn using "pROC" software package to evaluate the prediction efficiency of the model.
Results: The most efficacious model at predicting SLN (+) in breast cancer patients with different molecular subtypes and SLN metastasis was the model based on the imageological characteristics of fat inhibition, and T2-weighted imaging (T2WI), T1-weighted imaging + C (T1WI-C), and DWI combined sequences at the tumor + 2 mm periphery. AUC (sensitivity, specificity) of the validation group were 0.831 (0.856, 0.891), 0.832 (0.660, 0.877), 0.801 (0.772, 0.765), 0.904 (0.769, 0.873), and 0.819 (0.810, 0.500) respectively when the tumor was 2 mm around the tumor.
Conclusions: The imaging features extracted from multi-parameter DWI, T1WI+C, and T2WI in breast cancer have certain value at predicting different molecular types and SLNs of breast cancer.
Keywords: Clinical imaging; breast cancer; molecular typing; sentinel lymph nodes (SLNs).
2022 Annals of Translational Medicine. All rights reserved.