Search

Our paper proposes a method to measure lung parenchyma parameters from pulmonary window computed tomography images based on ResU-Net model including the CT value, the density, the lung volume, and the surface area of the lungs of healthy rats, to help promote the quantitative analysis of lung parenchyma parameters of rats in medical respiratory researches. Through the analysis of the lung parenchyma parameters of the control group and the treatment group, the law of change among the lung parenchyma parameters is given in our paper. After comparing and analyzing the lung parenchyma parameter CT value and the density of the two groups, it is discovered that the lung parenchyma parameter CT value and the density significantly increase in the treatment group which is after continuously inhaling the nebulization of contrast agents. The change of the lung volume with the surface area in both two groups conforms to the law of lung changes during breathing. The relationship between the lung volume and the CT value or the density is analyzed and it is concluded that the lung volume is negatively correlated with the CT value or the density.

Deep learning (DL) has emerged as a powerful image processing technique that learns the features of the data and produces state-of-the-art prediction results. The decade from 2010 to 2020 is a real revival of DL, which has come to a turning point in history. In image classification, many deep learning networks have been proposed by scholars, and each of them has its own strengthness. It is very important and efficient for the researchers and the developers to know the performance of these networks, especially for the beginners, so as to give them a transplant instruction by an objective evaluation index. In this paper, we constructed different data sets from three aspects, texture, shape, and measurement scale to test the performance of nine mainstream image classification networks. Cross-contrast experiments were performed to analyze the sensitivity of factors which influence the stability of image classification networks. Experimental results shown that in the 27 image datasets generated by the three image factors, the classification performance of AlexNet, GoogleNet, VggNet, and DenseNet is better. The perfomance comparison of these networks are showed and discussed in details. Code and pretrained models are available at https://github.com/liqiang12689/image-classification-finall .

To understand the influence of the text representation of vertigo interrogation texts on classification results, our paper proposes a fusion strategy of text representation combination and constructs a new stacking model fusion. Vertigo interrogation texts of the four typical vertigo diseases are used to make classification under the different text representation. Results show that the classification effect of proposed stacking model fusion with proposed text representation combinations is superior to that of the gradient boosting decision tree, support vector machine, and naive Bayes with the simple text representation. Results also show that the proposed text representation combination of word2vec, GloVe and one-hot is the best overall classification effect among other text representation combinations based on our proposed stacking model fusion. Therefore, we conclude that the proposed fusion strategy of text representation combination and stacking model fusion can improve the classification effect of text representation of vertigo interrogation texts.

As an important branch of medical robots, surgical robots have attracted the attention of studies in China and abroad, and these studies also have carried out many explorations and researches on them. Based on the current research results, this paper summarizes the development status of surgical robots and the development history of four surgical robots which are neurosurgery surgical robots, orthopedic surgical robots, endoscopic surgical robots and vascular intrusive surgical robots. From three aspects which include the structure design, the virtual simulation system and the control factor of the four surgical robots, this paper analyzes the key technology of surgical robots and finally predict and prospect the future development of the surgical robots.

We can observe the entire lung by the chest image which can acquire from the computed tomography device, but it is difficult to observe five pulmonary lobes separately, because the five pulmonary lobes are closely attached to each other. The spatial distribution of the chronic obstructive pulmonary disease on left and right lungs reflects the severity of airflow limitation. Traditional methods of locating chronic obstructive pulmonary diseases often require identification of the entire lung region, including left and right lungs. If both left lung and right lung are used to detect the spatial distribution of the chronic obstructive pulmonary disease, however some regions will be not regions of interest. Therefore, our paper essentially proposes a method for extracting a single pulmonary lobe with Hounsfield units from computed tomography pulmonary images for the diagnosis or the teaching. After abstracting five pulmonary lobes we can select the regions of interest such as a certain pulmonary lobe, then we can observe and/or analyze the selected certain pulmonary lobe, which is the region of Interest, such as locating the chronic obstructive pulmonary disease.

The combination of artificial intelligence technology and medical science has inspired the emergence of medical robots with novel functions that use new materials and have a neoteric appearance. However, the diversity of medical robots causes confusion regarding their classification. In this paper, we review the concepts pertinent to major classification methods and development status of medical robots. We survey the classification methods according to the appearance, function, and application of medical robots. The difficulties surrounding classification methods that arose are discussed, for example, (1) it is difficult to make a simple distinction among existing types of medical robots; (2) classification is important to provide sufficient applicability to the existing and upcoming medical robots; (3) future medical robots may destroy the stability of the classification framework. To solve these problems, we proposed an innovative multilevel classification strategy for medical robots. According to the main classification method, the medical robots were divided into four major categories—surgical, rehabilitation, medical assistant, and hospital service robots—and personalized classifications for each major category were proposed in secondary classifications. The technologies currently available or in development for surgical robots and rehabilitation robots are discussed with great emphasis. The technical preferences of surgical robots in the different departments and the rehabilitation robots in the variant application scenes are perceived, by which the necessity of further classification of the surgical robots and the rehabilitation robots is shown and the secondary classification strategy for surgical robots and rehabilitation robots is provided. Our results show that the distinctive features of surgical robots and rehabilitation robots can be highlighted and that the communication between professionals in the same and other fields can be improved.