Classification of H&E Stained Liver Histopathology Images Using Ensemble Learning Techniques for Detection of the Level of Malignancy of Hepatocellular Carcinoma (HCC)

Hepatocellular carcinoma (HCC) is one of the most common types of primary liver cancer and a leading cause of cancer-related deaths worldwide. Diagnosis of the HCC using H&E stained liver histopathology images is a challenging task mainly because of the time-consuming and error-prone manual examination even when performed by skilled pathologists. Deep learning has revolutionized medical diagnosis by providing accurate and economical computer-aided diagnosis solutions. The classification of histopathology images with algorithms such as convolutional neural networks has shown promising results in recent studies. To overcome the limitations of previous studies and provide a more efficient solution, this study proposes a novel deep learning-based ensemble model. The publicly available TCGA-LIHC dataset, consisting of three cancer types based on severity level, is pre-processed and then further used for model development and evaluation. Four pre-trained CNN models AlexNet, VGG16, Inception-v3, and ResNet50 that follow transfer learning approaches are used to build the ensemble model. The proposed ensemble model outperforms the state-of-the-art CNN architectures in terms of sensitivity, specificity, F1-score, accuracy, and area under the curve (AUC). The proposed model when compared with previously published models for the same TCGA-LIHC dataset, shows a better HCC prediction accuracy than most. In conclusion, we affirm that the proposed ensemble deep learning model is able to classify H&E stained histopathology images better than the state-of-the-art CNN architectures, ultimately providing a time-saving and precise solution to the HCC classification task.