PneuNet: deep learning for COVID-19 pneumonia diagnosis on chest X-ray image analysis using Vision Transformer.

A long-standing challenge in pneumonia diagnosis is recognizing the pathological lung texture, especially the ground-glass appearance pathological texture. One main difficulty lies in precisely extracting and recognizing the pathological features. The patients, especially those with mild symptoms, show very little difference in lung texture, neither conventional computer vision methods nor convolutional neural networks perform well on pneumonia diagnosis based on chest X-ray (CXR) images. In the meanwhile, the Coronavirus Disease 2019 (COVID-19) pandemic continues wreaking havoc around the world, where quick and accurate diagnosis backed by CXR images is in high demand. Rather than simply recognizing the patterns, extracting feature maps from the original CXR image is what we need in the classification process. Thus, we propose a Vision Transformer (VIT)-based model called PneuNet to make an accurate diagnosis backed by channel-based attention through X-ray images of the lung, where multi-head attention is applied on channel patches rather than feature patches. The techniques presented in this paper are oriented toward the medical application of deep neural networks and VIT. Extensive experiment results show that our method can reach 94.96% accuracy in the three-categories classification problem on the test set, which outperforms previous deep learning models.

Research on the Evolution of Shield Segment Cracks Based on Acoustic Emission and CMOD.

In order to explore the cracking law and failure characteristics of segments, a model test of shield segment cracking was conducted. The microscopic and macroscopic crack evolution process of the segment is studied by using acoustic emission detection technology and crack opening displacement (CMOD). According to the acoustic emission signal and CMOD, characteristics generated in the process of segment cracking, in the form of numerical value, the evolution characteristics of each stage of segment cracking are directly reflected. Based on acoustic emission energy and CMOD, the segment cracking damage model was established to determine the segment fracture damage degree. The result shows that segment cracking can be divided into three stages, and the acoustic emission detection results and CMOD have different degrees of change in each cracking stage. This proves that both the acoustic emission acquisition results and CMOD can be used as evaluation indicators of damage degree. Acoustic emission can accurately identify the crack evolution process, and the yield strengthening is an important stage of crack damage evolution. The damage data points in this stage account for 76.83% of all the damage data points, the occurrence rate of damage data points is 0.225 s, and the density of data points in the damaged area is 3.219 × 10-4 mm3, which is larger than the other two stages. The segment cracking damage model can effectively reflect the segment cracking degree and provide a reference for the actual segment cracking assessment.