Multi-attention Network for Thoracic Disease Classification and Localization

Ma, Yang; Zhou, Qiuhao; Chen, Xuesong; Lu, Haihua; Zhao, Yong

doi:10.1109/icassp.2019.8682952

Cited by 19 publications

(17 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This section will introduce related deep learning-based thorax disease classification methods since our approach is CNN based. According to different learning strategies, current deep learning-based methods can be broadly grouped into three categories including: (1) thorax disease classification using global information [ 9 , 22 , 23 , 24 , 25 , 26 , 27 , 28 ], (2) thorax disease classification using global and local information [ 12 , 13 ] and (3) thorax disease classification using visual attention [ 16 , 17 , 18 , 19 , 20 ].…”

Section: Related Workmentioning

confidence: 99%

“…Recently, a number of methods use visual attention to optimize deep neural networks for thorax disease classification [ 16 , 17 , 18 , 19 , 20 ]. For example, Sorkhei et al [ 19 ] add a space attention module on top of a pre-trained ResNet to capture global context features, which are then combined with original feature maps (local features).…”

Section: Related Workmentioning

confidence: 99%

“…They optimize each attention module by corresponding disease label and thus each attention map consists of feature representations that are related to each category. Ma et al [ 17 ] propose a multi-attention learning framework for comprehensive thoracic disease classification and localization, which consists of a feature attention module, a space attention module and a hard example attention module. Specifically, the feature attention module is a squeeze-and-excitation structure, which is equipped after each residual block of ResNet101 to refine the extracted feature maps at multiple resolutions.…”

Section: Related Workmentioning

confidence: 99%

“…Due to its good potential to train a better deep model, visual attention has been widely used for various computer vision tasks, i.e., image classification [ 14 ] and person re-identification [ 15 ], and has brought significant performance improvements. Visual attention also has been applied in thorax disease classification in recent years, but most existing methods [ 16 , 17 , 18 , 19 , 20 ] learn global attention maps from whole CXR images only where local informative cues lying around specific organs are often suppressed. To refine the learned global feature, several works [ 16 , 17 , 19 ] have been reported to either use single attention module or use multiple attention modules to re-weight the learned feature representations of backbone network.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Part-Aware Mask-Guided Attention for Thorax Disease Classification

Zhang

Yang

Luo

et al. 2021

Entropy

View full text Add to dashboard Cite

Thorax disease classification is a challenging task due to complex pathologies and subtle texture changes, etc. It has been extensively studied for years largely because of its wide application in computer-aided diagnosis. Most existing methods directly learn global feature representations from whole Chest X-ray (CXR) images, without considering in depth the richer visual cues lying around informative local regions. Thus, these methods often produce sub-optimal thorax disease classification performance because they ignore the very informative pathological changes around organs. In this paper, we propose a novel Part-Aware Mask-Guided Attention Network (PMGAN) that learns complementary global and local feature representations from all-organ region and multiple single-organ regions simultaneously for thorax disease classification. Specifically, multiple innovative soft attention modules are designed to progressively guide feature learning toward the global informative regions of whole CXR image. A mask-guided attention module is designed to further search for informative regions and visual cues within the all-organ or single-organ images, where attention is elegantly regularized by automatically generated organ masks and without introducing computation during the inference stage. In addition, a multi-task learning strategy is designed, which effectively maximizes the learning of complementary local and global representations. The proposed PMGAN has been evaluated on the ChestX-ray14 dataset and the experimental results demonstrate its superior thorax disease classification performance against the state-of-the-art methods.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Part-Aware Mask-Guided Attention for Thorax Disease Classification

Zhang

Yang

Luo

et al. 2021

Entropy

View full text Add to dashboard Cite

show abstract

“…Deep learning have shown promise in the field of medical image analysis [6]- [8], [14], [28], [29], especially in multilabel chest X-rays recognition [17], [18], [21], [30]. Yao et al [18] proposed a method combining Long-short Term Memory Network (LSTM) and DenseNet network [31] to predict thoracic disease through label correlation.…”

Section: A Deep Learning For Multi-label Chest X-rays Recognitionmentioning

confidence: 99%

Learning to Recognize Thoracic Disease in Chest X-Rays With Knowledge-Guided Deep Zoom Neural Networks

et al. 2020

View full text Add to dashboard Cite

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.

show abstract