Gastric Cancer Invasion Depth

A VGG-16-based AI model initially designed to predict invasion depth in early gastric cancer (EGC) using static images was found to be limited in capturing real-time spatio-temporal information. To address this, a video classifier (VC) was developed, which outperformed the image classifier (IC) with higher sensitivity, specificity, and accuracy (82.3%, 85.8%, and 83.7%, respectively). The VC also showed more consistent predictions, making it more suitable for real-world clinical applications in predicting EGC invasion depth.