Computed tomography 3D reconstruction and texture analysis for evaluating the efficacy of neoadjuvant chemotherapy in advanced gastric cancer

Figure 1 Column line chart prediction model. This is a scoring system consisting of three main evaluation indicators: Thickness change rate, gastric wall motility, and disappearance. Each indicator has a score range of 0-90 points, with a total score ranging from 0 to 320 points. The system displays a tumor regression grade incidence rate indicator at the bottom, ranging from 0.4 to 0.95. This scoring system can be used to evaluate medical conditions or treatment effectiveness. TRG: Tumor regression grade.

Figure 2 Receiver operating characteristic curve of volume change rate and computed tomography subjective efficacy model. This receiver operating characteristic curve shows a comparison of two indicators: Predicted probability: Area under the curve (AUC) = 0.81966, volume change rate: AUC = 0.72306. AUC: Area under the curve.

Figure 3 Calibration curve of prediction model. This is a calibration curve used to evaluate model prediction accuracy. The graph contains three curves: A prediction curve (dashed line), a calibrated line curve (red dotted line), and an ideal curve (solid line). Both the x-axis (predicted probability) and y-axis (observed probability) range from 0.4 to 0.9. Ideally, the prediction curve should align closely with the ideal curve, indicating good predictive accuracy of the model.

Figure 4 Analysis of decision curve analysis curve of prediction model. The graph compares the net benefit of three treatment strategies: Prediction model, treat all, and treat none. Before a risk threshold of 0.5, treating all patients shows the best results; after that, the prediction model performs better. The treat-none strategy maintains near-zero benefit throughout. This helps select the optimal treatment approach based on patient risk levels.