GI-Net: Anomalies Classification in Gastrointestinal Tract through Endoscopic Imagery with Deep Learning

Recently, gastrointestinal(GI) tract disease diagnosis through endoscopic image classification is an active research area in the biomedical field. Several GI tract disease classification methods based on image processing and machine learning techniques have been proposed by diverse research groups in the recent past. However, yet effective and comprehensive deep ensemble neural network-based classification model is not available in the literature. In this research work, we propose to use an ensemble of deep features as a single feature vector by combining pre trained DenseNet-201, ResNet-18, and VGG-16 CNN models as the feature extractors followed by a global average pooling (GAP) layer to predict eight-class anomalies of the digestive tract diseases. Our results show a promising accuracy of over 97% which is a remarkable performance with respect to the state-of-the-art approaches. We analyzed how prominent CNN architectures that have appeared recently (DenseNet, ResNet, Xception, InceptionV3, InceptionResNetV2, and VGG) that can be used for the task of transfer learning. Furthermore, we describe a technique of reducing processing time and memory consumption while preserving the accuracy of the classification model by using feature extraction based on SVD.