Artificial Intelligence-Based Tissue Phenotyping in Colorectal Cancer Histopathology Using Visual and Semantic Features Aggregation

Tissue phenotyping of the tumor microenvironment has a decisive role in digital profiling intra-tumor heterogeneity, epigenetics, and progression cancer. Most existing methods for tissue often rely on time-consuming error-prone manual procedures. Recently, with advent advanced technologies, these procedures have been automated using artificial intelligence techniques. In this paper, novel deep histology heterogeneous feature aggregation network (HHFA-Net) is proposed based visual semantic information fusion detection phenotypes colorectal cancer (CRC). We adopted tested various data augmentation techniques to avoid computationally expensive stain normalization handle limited imbalanced problems. Three publicly available datasets are used experiments: CRC (CRC-TP), (CRCH), colon (CCH). The HHFA-Net achieves higher accuracies than state-of-the-art histopathology images.