| Breast cancer is the most common cancer in women,and neoadjuvant chemotherapy is the best treatment option for early-stage breast cancer.Due to the high heterogeneity of breast tumors and the differences in the physical conditions of different people,the curative effect and prognosis of different patients after chemotherapy are also different,and the corresponding follow-up treatment options are also different.Studies have shown that the survival rate of patients with effective chemotherapy is significantly better than that of patients with ineffective chemotherapy,and the regression model is also related to patient survival,downstaging and tumor shrinkage.Dynamically enhanced MRI has the advantages of non-invasiveness,high specificity,and high sensitivity.If the image data before chemotherapy can be used to predict the shrinkage pattern and chemotherapy efficacy markers of patients after chemotherapy,it can help doctors to carry out precise clinical treatment at an earlier stage,improve the treatment plan and improve the prognosis of patients.However,studies based on radiomics do not have biological significance.From the perspective of genomics,the relationship between gene expression and chemotherapy efficacy and the prediction results of shrinkage patterns can be evaluated,thereby providing potential interpretation of molecular biology.The main content of the paper is as follows:(1)Multi-task chemotherapy efficacy and shrinkage model prediction research based on radiomics.First extract the features of the region of interest,and after screening the features with the highest frequency,use the same features to establish a multitask model to predict chemotherapy efficacy and shrinkage pattern.The results show that the multi-task model has good predictive performance in the test set(chemotherapy efficacy AUC=0.760,shrinkage pattern AUC=0.886).(2)The application of predictive labels for chemotherapy efficacy and shrinkage pattern in survival analysis.After feature extraction of the TCGA-ISPY1 dataset,the multi-task model can be used to predict the efficacy and shrinkage pattern of patients after chemotherapy.For patients with different labels,survival analysis showed significant differences in survival: patients with effective chemotherapy had the best survival,and patients with concentric shrinkage had a more ideal prognosis than those with nonconcentric shrinkage.The prognosis of patients with ineffective chemotherapy could be further judged by the shrinkage pattern.(3)Interpretable and reproducible studies of chemotherapy efficacy and shrinkage patterns in molecular biology.Through the KEGG gene enrichment analysis of the differentially expressed genes in GEO-ISPY1 that are highly correlated with the curative effect-shrinkage group,the signaling pathways that are highly correlated with the group were obtained,and the survival differences of different groups were preliminarily determined from the perspective of molecular biology.explanatory inquiry.The results show that the IL-17 signaling pathway,estrogen signaling pathway,and endoplasmic reticulum protein processing pathway may have potential connections with the survival of different groups,and genes such as BCL2 and ATF6 B frequently appear in the above pathways,and their expression values are positive and negative with the grouping correlation.Finally,it is found that the prediction model and survival model of some image features in different data sets have a consistent trend,which proves the repeatability value of features such as sphericity.The chemotherapy efficacy and shrinkage pattern predicted by dynamic enhanced MRI images before neoadjuvant chemotherapy have value as prognostic markers,which can provide reference value for doctors in clinical diagnosis,prognosis,and precise treatment during early neoadjuvant chemotherapy.At the same time,through genomics analysis,the study found that the survival rates of different prediction groups may be affected by some signaling pathways,providing biological support for the above conclusions. |
PDF Full Text Request