Automated Detection of Multiple Lesions on Chest X-ray Images: Classification Using a Neural Network Technique with Association-Specific Contexts

Abstract

Automated detection of lung lesions on Chest X-ray images shows good performance to reduce lung cancer mortality. However, it is difficult to detect multiple lesions of single image well and truly, and additional efforts are needed to improve diagnostic efficiency and quality. In this paper, a multi-label classification model combining attention-based neural networks and association-specific contexts is proposed for the detection of multiple lesions on chest X-ray images. A convolutional neural network and a long short-term memory network are first aligned by an attention mechanism to take advantage of both image and text information for the detection, called CNN-ATTENTION-LSTM (CAL) network. In addition, a mining method of implicit association strength to obtain an association network of chest lesions (CLA) network is designed to guide the training of CAL network. The CLA network provides possible clinical relationships between lesions to help the CAL network obtain better predictions. Experimental results on ChestX-ray14 dataset show that our method outperforms some state-of-the-art models under the metrics of area under curve (AUC), precision, recall, and F-score and achieves up to 85.4% in the case of atelectasis and infiltration. It indicates that the method may be useful in the computer-aided detection of multiple lesions on chest X-ray images.

Detection of sitting posture using hierarchical image composition and deep learning

“…Reducing the dimensionality twice leaves us with 128 features, each of 160 × 120 px resolution. At this stage, we use Long Short Term Memory (LSTM) convolutional block ( Xu et al, 2020 ; Li et al, 2020 ), which is tasked to extract 32 most useful features in the entire sequence. For our main neural network backbone, we use MobileNetV2 , which is the extension of MobileNet , for it has show to achieve great results in predictive capabilities ( Howard et al, 2017 ; Zhou et al, 2019 ), however, the architecture itself is relatively light-weight for it is designed to be used in low power devices such as mobile devices, unlike for example, YOLOV3 , which while having impressive recall results ( Redmon & Farhadi, 2018 ), is much more complex and has a substantially poorer performance.…”

Deep learning modified neural networks with chicken swarm optimization-based lungs disease detection and severity classification

“…The x-ray picture size adjustment is one of the most critical data preprocessing processes because CNNs have certain input requirements. 25 DenseNet201 requires a 299 × 299 × 3 image, whereas exception and InceptionV3 just require a 224 × 224 × 3 image. It required a (224 × 224 × 3) shape to be carved from the x-ray images.…”

“…The chest x-ray pictures in the data collection must first be preprocessed before they can be analyzed. The x-ray picture size adjustment is one of the most critical data preprocessing processes because CNNs have certain input requirements 25 . DenseNet201 requires a 299×299×3 image, whereas exception and InceptionV3 just require a 224×224×3 image.…”

A Study on the Pulmonary Diseases using Deep Learning

“…For the diagnosis of many malignancies in chest X-ray images, ShuaijingXu and coworkers [12] present an attention-driven ensemble learningand association context model. The CNN-ATTENTION-LSTM (CAL) network at first integrates a CNN model with a long short-term memory mechanism to identify items in both texts and images [13,14].…”