Quantitative analysis of chest MRI images for benign malignant diagnosis of pulmonary solid nodules

Background In this study, we developed and validated machine learning (ML) models by combining radiomic features extracted from magnetic resonance imaging (MRI) with clinicopathological factors to assess pulmonary nodule classification for benign malignant diagnosis. Methods A total of 333 consecutive patients nodules (233 in the training cohort 100 validation cohort) were enrolled. 2,824 MRI images (CE T1w T2w). Logistic regression (LR), Naïve Bayes (NB), support vector (SVM), random forest (RF), extreme gradient boosting (XGBoost) classifiers used build predictive models, a radiomics score (Rad-score) was obtained each patient after applying best prediction model. Clinical Rad-scores jointly nomogram model based on multivariate logistic analysis, diagnostic performance five evaluated using area under receiver operating characteristic curve (AUC). Results 161 women (48.35%) 172 men (51.65%) Six important selected 2,145 CE T2w images. The XGBoost classifier achieved highest discrimination AUCs 0.901, 0.906, 0.851 training, validation, test cohorts, respectively. improved AUC values 0.918, 0.912, 0.877 Conclusion ML demonstrated good XGBoost, which superior that other four models. higher addition clinical information.