DCE-MRI影像组学特征对浸润性乳腺癌前哨淋巴结转移负荷的预测价值OA

Objective To explore the predictive value of DCE-MRI radiomics features for sentinel lymph node(SLN)metastasis burden in invasive breast cancer.Methods From December 2022 to December 2024,186 breast cancer patients treated in our hospital were stratified and divided into a training set(n=130)and a validation set(n=56)in a 7∶3 ratio.Based on postoperative pathological lymph node metastasis status,the patients were further stratified into a high metastatic burden group(n=99)and a low metastatic burden group(n=87).Least Absolute Shrinkage and Selection Operator(LASSO)regression was employed to screen omics features.Multivariable Logistic analysis was conducted to investigate factors influencing sentinel lymph node metastasis,and a predictive model for sentinel lymph node metastasis was established to evaluate the predictive performance of each feature.Results Significant differences were observed between the high metastatic burden groupand the low metastatic burden group in terms of patient age,family history of cancer,and pathological type(P<0.05).DCE-MRI imaging data revealed statistically significant differences in the maximum tumor diameter,the proportion with lobulation,the proportion with irregular shapes,the proportion with unclear boundaries,the proportion with obvious background parenchymal enhancement,and the elastic strain rate in the high metastatic burden group(P<0.05).Multivariate analysis indicated that age,family history of cancer,pathological type,maximum tumor diameter,presence of lobulation,shape,boundary,background parenchymal enhancement,ADC value,elastic strain rate,DCE_wavelet-HH_glrim_ShortRunHighGrayLevelEmphasis,wavelet_glcm_wavelet-1lh-imc1,original_shape_leastaxislength,original_shape_majoraxislength,DCE_exponential_ngtdm_Complexity,and DCE_wavelet-LH_glcm_Clustershade were independent factors influencing sentinel lymph node metastasis.The predictive model established based on these factors had a sensitivity of 76.8%and a specificity of 83.6%on the Receiver Operating Characteristic(ROC)curve.The prediction model constructed based on influencing factors demonstrated an AUC of 0.853(95%CI:0.792-0.897)before internal validation and 0.867(95%CI:0.819-0.907)after validation,with sensitivities of 86.93%and 87.89%,and specificities of 91.05%and 93.82%,respectively.Conclusions DCE-MRI radiomics features have significant application value in predicting SLN metastasis burden in invasive breast cancer.This method can provide reference information for clinical diagnosis,subsequent treatment,and prognosis assessment.