Multiregional radiomics features from multiparametric MRI for prediction of lymphovascular invasion in rectal cancer

The presence of lymphovascular invasion (LVI) is thought to indicates an increased risk for progressive disease in rectal cancers according to the National Comprehensive Cancer Network (NCCN) Guidelines. Here, we developed and validated a radiomics model for prediction of LVI in rectal cancer based on pre-treatment MRI. The Ridge Classifier have the best prediction accuracy score( 73.3%), its specificity, sensitivity and F1 score are 83.9%, 59.4 % and 65.1 %, respectively. So, the radiomics features from MRI of rectal cancer is a useful tool for predicting LVI preoperatively and has marked discrimination accuracy.