Automatic and accurate thorax disease diagnosis in Chest X-ray (CXR) image plays an essential role in clinical assist analysis. However, due to its imaging noise regions and the similarity of visual features between diseases and their surroundings, the precise analysis of thoracic disease becomes a challenging problem. In this study, we propose a novel knowledge-guided deep zoom neural network (KGZNet) which is a data-driven model. Our approach leverage prior medical knowledge to guide its training process, due to thoracic diseases typically limit within the lung regions. Also, we utilized weaklysupervised learning (WSL) to search for finer regions without using annotated samples. Learning on each scale consists of a classification sub-network. The KGZNet starts from global images, and iteratively generates discriminative part from coarse to fine; while a finer scale sub-network takes as input an amplified attended discriminative region from previous scales in a recurrent way. Specifically, we first train a robust modified U-Net model of lung segmentation and capture the lung area from the original CXR image through the Lung Region Generator. Then, guided by the attention heatmap, we obtain a finer discriminative lesion region from the lung region images by the Lesion Region Generator. Lastly, the most discriminative features knowledge is fused, and the complementary features information is learned for final disease prediction. Extensive experiments demonstrate that our method can effectively leverage discriminative region information, and significantly outperforms the other state-of-the-art methods in the thoracic disease recognition task.

Recently, clustering algorithms based on deep AutoEncoder attract lots of attention due to their excellent clustering performance. On the other hand, the success of PCA-Kmeans and spectral clustering corroborates that the orthogonality of embedding is beneficial to increase the clustering accuracy. In this paper, we propose a novel dimensional reduction model, called Orthogonal AutoEncoder (OAE), which encourages the orthogonality of the learned embedding. Furthermore, we propose a joint deep Clustering framework based on Orthogonal AutoEncoder (COAE), and this new framework is capable of extracting the latent embedding and predicting the clustering assignment simultaneously. The COAE stacks a fully connected clustering layer on top of the OAE, where the activation function of the clustering layer is the multinomial logistic regression function. The loss function of the COAE contains two terms: the reconstruction loss and the clustering-oriented loss. The first one is a data-dependent term in order to prevent overfitting. The other one is the cross entropy between the predicted assignment and the auxiliary target distribution. The network parameters of the COAE can be effectively updated by the mini-batch stochastic gradient descent algorithm and the back-propagation approach. The experiments on benchmark datasets empirically demonstrate that the COAE can achieve superior or competitive clustering performance as state-of-the-art deep clustering frameworks.

No abstract