Deep learning model combined with computed tomography features to preoperatively predicting the risk stratification of gastrointestinal stromal tumors

Figure 1 Study population flowchart. CT: Computed tomography; GIST: Gastrointestinal stromal tumors.

Figure 2 Overall gastrointestinal stromal tumor risk stratification framework. CT: Computed tomography; MLP: Multilayer perceptron; GAP: Global average pooling.

Figure 3 Schematic diagram of feature extraction process. BN: Batch normalization; RELU: Rectified linear unit.

Figure 4 Receiver operating characteristic curves for the deep learning model and combined model. A: Receiver operating characteristic curve (ROC) for the deep learning model in the validation cohort; B: ROC curve for the combined model in the validation cohort. ROC: Receiver operating characteristic; AUC: Area under the curve.

Figure 5 Confusion matrix for the deep learning model and combined model for gastrointestinal stromal tumors risk stratification. A: Confusion matrix for the deep learning model; B: Confusion matrix for the combined model.