In this paper, we investigate the problem of predicting the histopathological findings of gastric cancer (GC) frompreoperative CT image. Unlike most existing classification systems assess the global imaging phenotype of tissuesdirectly, we formulate the problem as a generalized multi-instance learning (GMIL) task and design a deep GMILframework to address it. Specifically, the proposed framework aims at training a powerful convolutional neural network(CNN) which is able to discriminate the informative patches from the neighbor confusing patches and yield accuratepatient-level classification. To achieve this, we firstly train a CNN for coarse patch-level classification in a GMILmanner to develop several groups which contain the informative patches for each histopathological category, theintra-tumor ambiguous patches, and the extra-tumor irrelative patches respectively. Then we modify the fully-connectedlayer to introduce the latter two classes of patches and retrain the CNN model. In the inference stage, patient-levelclassification is implemented based on the group of candidate informative patches automatically recognized by the model.To evaluate the performance and generalizability of our approach, we successively apply it to predict two kinds ofhistopathological findings (differentiation degree [two categories] and Lauren classification [three categories]) on adataset including 433 GC patients with venous phase contrast-enhanced CT scans. Experimental results reveal that ourdeep GMIL model has a powerful predictive ability with accuracies of 0.815 and 0.731 in the two applicationsrespectively, and it significantly outperforms the standard CNN model and the traditional texture-based model (morethan 14% and 17% accuracy increase).
展开▼
