Differentiation of gastric schwannomas from gastrointestinal stromal tumors by CT using machine learning

摘要: OBJECTIVE To identify schwannomas from gastrointestinal stromal tumors (GISTs) by CT features using Logistic Regression (LR), Decision Trees (DT), Random Forest (RF), and Gradient Boosting Tree (GBDT). METHODS This study enrolled 49 patients with 139 GISTs proven pathology. P < 0.1 derived univariate analysis were inputted to four models. Five machine learning (ML) versions, multivariate analysis, radiologists' subjective diagnostic performance compared evaluate diagnosis performance of all the traditional advanced methods. RESULTS The as follows: (1) attenuation value unenhancement phase (CTU), (2) portal venous enhancement (CTV), (3) degree in (DEPP), (4) minus arterial (CTV-CTA), (5) enhanced potentiality (EP), (6) location, (7) contour, (8) growth pattern, (9) necrosis, (10) surface ulceration, (11) enlarged lymph node (LN). LR (M1), RF, DT, GBDT models contained above 11 variables, while (M2) was developed six most predictive variables (M1). model AUC 0.967 test dataset thought be optimal differentiating two tumors. Location gastric body, exophytic mixed lack necrosis nodes, larger EP important suggestive schwannomas. CONCLUSION provided potency among other ML on differentiation GISTs.