Prognostic Prediction of Gastric Cancer Based on Ensemble Deep Learning of Pathological Images

The analysis of pathological images is of great significance for the diagnosis and prognosis of gastric cancer. However, its clinical application still faces challenges such as low consistency of visual reading and large differences in multi-resolution images. To address these issues, a prognostic prediction method of gastric cancer based on ensemble deep learning of pathological images is proposed. First, the pathological images at different resolutions of patient are preprocessed by slicing and filtering. Then, the deep feature extraction and fusion of slices at different resolutions are carried out by using three deep learning methods, i.e., ResNet, MobileNetV3, and EfficientNetV2, respectively, which aims to obtain the single-resolution prediction results of individual classifier at patient level. Finally, a double-level ensemble strategy is used to fuse the prediction results of heterogeneous individual classifiers at different resolutions to obtain the patient-level prognostic prediction results. In the experiment, the pathological images of 250 gastric cancer patients are collected, and the prediction of distant metastases is used as an example for verification. The experimental results show that the prediction accuracy of the proposed method on the test set is 89.10%, the sensitivity is 89.57%, the specificity is 88.61%, and the Matthews correlation coefficient is 78.19%. Compared with the single-model prediction results, the prediction performance of the proposed method has been significantly improved, which can provide an important reference for the treatment and prognosis of gastric cancer patients.