Multicenter study demonstrates improved prediction of high-risk gastroesophageal varices using deep learning and two-dimensional shear wave elastography.
Key Points
The deep learning risk prediction model significantly outperforms traditional indicators for high-risk gastroesophageal varices.
Area under the curve values were 0.91 for gastroesophageal varices and 0.88 for high-risk gastroesophageal varices.
Analysis included liver stiffness measurement and spleen stiffness measurement images, enhancing prediction accuracy.
Spleen stiffness measurement images contributed more significantly to predictions than liver stiffness measurement images.