Ultrasound radiomics-based nomogram to predict lymphovascular invasion in invasive breast cancer: a multicenter, retrospective study

Objectives To develop and validate an ultrasound (US) radiomics-based nomogram for the preoperative prediction of the lymphovascular invasion (LVI) status in patients with invasive breast cancer (IBC). Materials and methods In this multicentre, retrospective study, 456 consecutive women were enrolled from three institutions. Institutions 1 and 2 were used to train ( n = 320) and test ( n = 136), and 130 patients from institution 3 were used for external validation. Radiomics features that reflected tumour information were derived from grey-scale US images. The least absolute shrinkage and selection operator and the maximum relevance minimum redundancy (mRMR) algorithm were used for feature selection and radiomics signature (RS) building. US radiomics-based nomogram was constructed by using multivariable logistic regression analysis. Predictive performance was assessed with the receiving operating characteristic curve, discrimination, and calibration. Results The nomogram based on clinico-ultrasonic features (menopausal status, US-reported lymph node status, posterior echo features) and RS yielded an optimal AUC of 0.88 (95% confidence interval [CI], 0.84–0.91), 0.89 (95% CI, 0.84–0.94) and 0.95 (95% CI, 0.92–0.99) in the training, internal and external validation cohort. The nomogram outperformed the clinico-ultrasonic and RS model ( p