Chinese Journal of Biomedical Engineering

Home

Editorial Board

Download

Quick Search

Office Online

Current Issue

Download Articles

Download

Links

2020 Vol. 39, No. 4 Published: 2020-08-20

Reviews Communications Regular Papers

Regular Papers

	385	Data-Driven Automatic Segmentation Algorithm for Trigeminal Nerve Fiber
		Jin Er, Feng Yuanjing, Zeng Qingrun, Chen Yukai, Huang Shengwei, Ruan Linhui
		DOI: 10.3969/j.issn.0258-8021.2020.04.001
		A common problem in the process of trigeminal nerve fiber tractography is artificial dependence,mainly including artificial rendering of region of interest (ROI) and manual screening of target fibers,which generally results in uncertainty and data errors. To ovecome this problem,a data-driven automatic trigeminal nerve fiber segmentation algorithm was proposed in this paper. A data-driven fiber clustering atlas was established based on the fiber data of several groups of brain samples,which automatically segmented the fiber data of new samples and directly obtained the trigeminal nerve fibers. In experiments,25 groups of healthy young individuals were selected as samples. Firstly,the brainstem was extracted by FSL software segmentation tool as ROI for deterministic fiber tracking. Secondly,a data-driven clustering atlas of fibers was created by multi-sample registration and spectral clustering of 20 groups of fibers. According to the tiny characteristics of trigeminal nerve,the trigeminal nerve fibers were labeled by secondary classification of brainstem fibers in the process of establishing fiber atlas. Finally,new sample data of 5 groups of healthy young people were selected,and their brainstem fiber data were automatically segmented using fiber atlas to obtain trigeminal nerve fiber bundles,and theweighted Dice coefficient between the results of automatic segmentation and manual segmentation of the same sample data was calculated. Results showed that the proposed method successfully segmented 5 sets of trigeminal nerve fiber bundles while the conventional manual method successfully identified 4 sets. The weighted Dice coefficients between the two results were 0.865,0.939,0.824,and 0.942. These results showed that this method can effectively avoid the influence of human factors,and greatly improve the work efficiency of neurosurgeons and cranial nerve researchers.
		2020 Vol. 39 (4): 385-393 [Abstract] ( 290 ) HTML (1 KB) PDF (8987 KB) ( 136 )

	394	2D/3D Medical Image Registration Using Convolutional Neural Network
		Chen Xiangqian, Guo Xiaoqing, Zhou Gang, Fan Yubo, Wang Yu
		DOI: 10.3969/j.issn.0258-8021.2020.04.002
		2D/3D registration is widely used in clinical diagnosis and surgical navigation planning,which can solve the problem of missing information in different dimensions of medical images and assist doctors to accurately locate patients′ lesions during surgery. The conventional 2D/3D registration method mainly relies on the gray level of the image for registration,but the registration process is very time consuming,which is not conducive to the clinical real-time requirements,and the registration process is easy to fall into the local optimum. This study proposed a deep learning approach to solve 2D/3D medical image registration problems. The method used a deep learning-based convolutional neural network to train the DRR and automatically learned image features to predict the parameters corresponding to the X-ray image to achieve registration. In the study,the human pelvis model bone was used as the experimental object. A total of 36,000 DRR images were generated as training sets,and 50 X-ray images of the model bone were collected by C arm for verification. Results showed that the test values for the three precision evaluation indicators of the correlation coefficient,normalized mutual information and Euclidean distance were 0.82±0.07,0.32±0.03,61.56±10.91 and the corresponding test values of the conventional 2D/3D algorithm were 0.79±0.07,0.29±0.03,37.92±7.24. These results meant the registration accuracy of deep learning algorithm was better than the conventional 2D/3D algorithm and there was no local optimal value for deep learning algorithm. Meanwhile,the registration time of deep learning was about 0.03 s,which was much lower than the time of conventional 2D/3D registration,which can satisfy the clinical demand for real-time registration. In the future,2D/3D deep learning registration research of clinical data will be further carried out.
		2020 Vol. 39 (4): 394-403 [Abstract] ( 888 ) HTML (1 KB) PDF (5192 KB) ( 652 )

	404	Assisted Diagnosis of Endoscopy Large Intestine Disease Based on Novel Feature and Bag of Feature Model
		Yang Jianjun, Chang Liping, Li Sheng, Zhu Tingwei, He Xiongxiong
		DOI: 10.3969/j.issn.0258-8021.2020.04.003
		Polyps and ulcerative colitis (UC) are common diseases of the large intestine. A large number of images generated during the endoscopy. To improve the diagnosis efficiency and accuracy,it is necessary to investigate the computer aided diagnosis system for the detection of colonscopy diseases. Considering the characteristics of endoscopy image,a novel color texture feature called histogram of local color difference was proposed in this paper,and used as the endoscopic image description by extracting local color difference histogram (LCDH) feature for each image patch in the feature extraction step. Combining with the bag-of-features model,local features were transformed into a higher-level image representation by using local-constrained linear coding and spatial pyramid matching. At last,SVM was used for classification. public Kvasir datasets were analyzed,and inferior images were deleted from original data and 5-fold cross validation was adopted. In the first experiment,the classification accuracy,sensitivity and specificity reached 97.88%,98.00% and 97.75% respectively for 800 normal samples and 800 disease samples;in the second experiment,1000 normal samples,770 polyp samples and 780 UC samples were adopted for multiple classification,the recognition rate of polyp and UC was 92.34% and 93.08% respectively. Experimental results showed that the proposed method possessed advantages both in accuracy and efficiency compared with the traditional method,which would be helpful for clinical diagnosis of intestinal diseases.
		2020 Vol. 39 (4): 404-412 [Abstract] ( 291 ) HTML (1 KB) PDF (6330 KB) ( 192 )

	413	Precancerous Diseases Classification Based on Fusion of Shallow and Deep Features
		Pan Yanqi, Chen Rui, Zhang Xu, Zhang Xinsen, Liu Jiquan, Hu Weiling, Duan Huilong, Si Jianmin
		DOI: 10.3969/j.issn.0258-8021.2020.04.004
		Precancerous disease recognition is of great significance in reducing the risk of gastric cancer. This paper proposed a method for identifying precancerous diseases based on the fusion of shallow and deep features of gastroscopic images. Firstly,according to properties of gastric images,75-dimensional shallow features were designed manually,including histogram features,texture features and higher order features. Secondly,based on the networks of Resnet and GoogLeNet,we added a fully connected layer before the output layer to extract the deep features of the images. To ensure consistent feature weights,the dimension of the added fully connected layer was 75. Finally,the shallow features were merged with deep features. Machine learning classifiers were used to identify three types of precancerous diseases,namely gastric polyps,gastric ulcers and gastric erosions. We collected 380 images for each disease,and 75% were used as training sets,the others were used as testing sets. We conducted experiments using traditional machine learning,deep learning and feature fusion proposed in this paper. Experiment results showed that the recognition accuracy of the feature fusion method proposed was as high as 95.18%,significantly better than that of traditional machine learning methods (74.12%) and deep learning methods (92.54%). This proposed method made full use of the shallow features and deep features to provide clinical decision support for doctors and assist in the diagnosis of precancerous diseases.
		2020 Vol. 39 (4): 413-421 [Abstract] ( 345 ) HTML (1 KB) PDF (6366 KB) ( 297 )

	422	Sleep Apnea Detection Based on Auto-encoder and Hidden Markov Model
		Qin Hengji, Liu Guanzheng
		DOI: 10.3969/j.issn.0258-8021.2020.04.005
		Obstructive sleep apnea (OSA) is prone to cardiovascular complications. As a gold standard for the diagnosis of sleep apnea,polysomnography is expensive and affects the sleep quality of patients. Because of the high coupling between heart and lung,electrocardiogram (ECG) signals are widely used in sleep apnea detection. However,most of the studies based on ECG signals focus on the design of artificial features,relying on the prior knowledge of experts. Methods based on deep learning can reduce human factors during feature extraction. In this study,we proposed a sleep apnea detection method based on auto-encoder and hidden Markov model (HMM). Firstly,a stacked sparse auto-encoder was used to perform semi-supervised feature learning directly from the RR interval sequence. Unsupervised learning was performed during the pre-training phase,and labels were then introduced for supervised learning during the fine-tuning phase. Then,a decision fusion classifier based on support vector machine (SVM) and artificial neural network (ANN) combined with HMM was constructed. HMM introduced the temporal dependence between segments. Decision fusion integrated the advantages between different classifiers. Experimental results based on the sleep data of 70 cases of all-night in PhysioNet′s apnea-ECG database showed that the accuracy,sensitivity and specificity of per-segment OSA detection was 84.7%,88.9% and 82.1% respectively,and per-subject detection accuracy was 100%. Compared with feature engineering,the feature extraction method based on auto-encoder could reduce the limitation of prior knowledge and make the feature extraction process more automatic and intelligent. In addition,compared with the single classifier,the decision fusion classifier not only improved the accuracy of per-segment OSA detection,but also alleviated the imbalance between sensitivity and specificity in detection results.
		2020 Vol. 39 (4): 422-431 [Abstract] ( 290 ) HTML (1 KB) PDF (1353 KB) ( 477 )

	432	Sleep Stage Classification Based on Heart Rate Variability Analysis and Model Performance Validation
		Zheng Jiewen, Zhang Yuezhou, Lan Ke, Liu Xiaoli, Zhang Zhengbo, Yu Mengsun
		DOI: 10.3969/j.issn.0258-8021.2020.04.006
		To conduct sleep stage classification and provide technical support for the application of wearable physiological monitoring technology in the field of chronic disease monitoring and management,a sleep stage classification algorithm based on heart rate variability (HRV) analysis and support vector machine (SVM) was developed. In order to ensure the quality of the training set,67 polysomnography (PSG) records were extracted by experts from the SHHS (Sleep Heart Health Study) PSG database for model training and internal validation. The sleep stages (wake,rapid eye movement and non-rapid eye movement) classified by EEG signals were used as labels to train the SVM model. Totally 86 features were derived from HRV analysis,including time domain,frequency domain and nonlinear domain. To test the generalization of the model,another 939 PSG records were further randomly extracted from the SHHS PSG database for model external validation. The accuracy of the 5-fold cross-validation on the training dataset of the 67 PSG records was 84.00%±1.33%,with a Kappa coefficient:0.70±0.03,and the accuracy of the algorithm on the 939 PSG records was 76.10%±10.8%,with a Kappa coefficient:0.57±0.15. The accuracy and the Kappa coefficient increased to 82.00%±5.6% and 0.67±0.14 when some records were excluded from the test dataset,including 110 records with abnormal RR intervals and 29 records with apparent abnormal sleep structures. These results showed that the model of heart rate variability analysis based sleep stage classification proposed in this paper exhibited a good performance,and the external validation by a dataset with large sample size demonstrated the generalization of the model.
		2020 Vol. 39 (4): 432-439 [Abstract] ( 399 ) HTML (1 KB) PDF (3443 KB) ( 686 )

	440	A Method Based on Bayesian Network to Retrieve Clinical Information Models: A Case Study of HL7 V3
		Huang Xiaoshuo, Yang Lin, Li Jiao
		DOI: 10.3969/j.issn.0258-8021.2020.04.007
		Reuse of clinical informationmodels (CIMs) is important for the interoperability of electronic health records. Retrieving and identifying reusable CIMs is an efficient way. In this study,we used the HL7 V3 2017 normative edition released by HL7.org as an example and applied an extended four-layer Bayesian network to represent this CIM. We enriched the hierarchicalmessage description (HMD) layer based on the simple Bayesian network,and calculated the semantic similarity between them. Thereafter,the reusable CIMs were identified through probability inferencing in the network. In the evaluation,we designed three retrieval tasks (“encounter appointment”,“laboratory result“and “patient entity”),and used MAP (mean average precision),AP (average precision),accuracy at cut-off point as evaluation metrics. Finally,we constructed a four-layer Bayesian network with 3 428 nodes and 22 646 edges,from top to bottom,the number of nodes was 2 177,422,422,407 for data element layer,HMD layer,duplicate HMD layer and message type layer respectively. Results showed that the value of MAP was 0.382,and the average accuracy at 3^rd,5^th and 10^th cut-off points was 77.8%,60.0% and 46.7% respectively. Our method was capable to retrieve general models and domain reusable models,as well as semantic related objects (such as “reschedule appointment notification“object in “encounter appointment” retrieval task). In summary,our method could help improve reusability of HL7 V3 CIMs as well as international standardization of clinical information,meanwhile,it could be useful for the optimization of other CIMs retrieval.
		2020 Vol. 39 (4): 440-448 [Abstract] ( 297 ) HTML (1 KB) PDF (843 KB) ( 183 )

	449	Electrochemical Algorithm Optimization Based on Multilayer Modified Electrode in Portable-Potentiostat System
		Xu Ying, Zhang Haijing, Dai Yan, Wang Xu, Chen Yangzi, Yang Yong
		DOI: 10.3969/j.issn.0258-8021.2020.04.008
		In order to calibrate the baseline drift of measurement curve caused by multi-layer modified electrode rapidly,a portable-potentiostat system was designed to detect E.coli with multilayer modified electrodes,and an electrochemical algorithm optimization in portable-potentiostat system was proposed in this paper. First,the algorithm used wavelet filter to remove noise by selecting the decomposition layer and threshold value. Then the baseline and peak-position calibration algorithms were used to remove the specificity error caused by intermediate modification of the electrode through calculation of peak position and base line slope. Results showed that the lower limit of system measurement reached the order of microamps (10^-6 A),the signal to noise ratio was improved by more than 30%,and the calibration error of peak-position was lower than 1.23%,achieving better denoising and calibration results. With the wide application of electrochemical biosensor technology based on complex modified materials in recent years,the algorithm could match the measured results with the multilayer modified electrode layer and made rapid qualitative longitudinal comparison and transverse prediction. The algorithm could also be used to predict the consistency of the results of electrochemical detection for a variety of microorganisms and trace characteristics,therefore showing great application potentials of fast electrochemical detection in the field of food safety.
		2020 Vol. 39 (4): 449-458 [Abstract] ( 221 ) HTML (1 KB) PDF (7687 KB) ( 47 )

	459	Finite Element Analysis of Posterior Atlantoaxial Fixationfor Type II Odontoid Process Fracture with High-Riding Vertebral Artery
		Dong Ziqiang, Zhao Gaiping, Bi Houhai, Zhao Qinghua, Wang Hongjie
		DOI: 10.3969/j.issn.0258-8021.2020.04.009
		To investigate the biomechanical characteristics on the treatment of type II odontoid process fracture with axial unilateral high-riding vertebral artery by two kinds of the combined posterior atlantoaxial fixation,the stability of the upper cervical vertebra and the stress distribution of the internal fixation implants translaminar screw,pedicle screws and C2 pars screw were analyzed. Based on the CT image data of type II odontoid process fracture of human cervical spine,combined with the finite element pre-processing software,according to the clinical operation plan,the posterior atlantoaxial fixation models of two combinations of upper cervical vertebra were established:1) unilateral axial translaminar screw + atlantoaxial pedicle screws fixation (C1PS-C2TL+PS);2) unilateral C2 pars screw + atlantoaxial pedicle screws fixation (C1PS-C2pars+PS). The range of motion and the stress distribution of internal fixation implants of two fixed models were analyzed under flexion and extension,lateral bending and rotation. In the two kinds of posterior atlantoaxial fixation on the treatment of type II odontoid process fracture with unilateral high-riding vertebral artery,the range of motion of the atlantoaxial joint of C1PS-C2TL+PS model decreased by 92.71%,91.28%,95.89% respectively under flexion and extension,lateral bending and rotation,and C1PS-C2pars+PS model decreased by 89.50%,94.77% and 92.72% respectively,all of which indicated that the stiffness of the fixed segments of vertebral body was significantly improved. In addition,the stress of C1PS-C2pars+PS model at the root of the axial screw and the lower part of the connecting rod was significantly concentrated under the flexion and extension conditions,with the maximum stress values of 179.9 and 167.6 MPa respectively,which was 55.1 and 52.2 MPa higher than those of C1PS- C2TL+PS model. The maximum stress values of C2pars screw under different conditions changed greatly,the maximum value was 123.7 MPa under flexion condition,which was 21.4 MPa higher than the maximum stress value of C2TL. In conclusion,the two fixed operations of C1PS-C2TL+PS and C1PS-C2pars+PS could effectively improve the stiffness of atlantoaxial for type II odontoid process fracture with high-riding vertebral artery,the former has better stability in flexion and extension and rotation,and the C1PS-C1TL+PS internal fixation is more reasonable in structure and stress distribution. These results provided theoretical basis for the study of internal fixation for type II odontoid process fracture with high-riding vertebral artery.
		2020 Vol. 39 (4): 459-465 [Abstract] ( 258 ) HTML (1 KB) PDF (3856 KB) ( 152 )

	466	HeLa Cells Uptake of Silica Coated Gold Nanorods and Intracellular Localization Based on Optical Imaging
		Sang Xiang, Wang Kexin, Xiao Shuanghuang, Yang Hongqin, Peng Yiru, Chen Jianling
		DOI: 10.3969/j.issn.0258-8021.2020.04.010
		Gold nanorods exhibit outstanding optical properties and high photothermal conversion efficiency,making them widely used in biomedical imaging and treatment. When gold nanorods are used in clinical treatments,their interaction with cells is a key issue. In this paper,the toxicity of silica coated gold nanorods to HeLa cells was studied. When HeLa cells were cultured with silica coated gold nanorods for different incubation time (4 h,8 h,12 h,24 h),the cellular uptake and intracellular distribution were observed by the two photon microscopy. We found out that the toxicity of silica coated gold nanorods to HeLa cells was incubation time and concentration dependent,and the serum in culture media could reduce the cytotoxicity. When silica coated gold nanorods were incubated with HeLa cells for 12 h,the viability of cells in serum-containing media was close to 100%,and in the serum free medium,the viability of cells was about 85% when the concentration of silica coated gold nanorods was1 250 μg/mL. After incubated for 24 h with 50 μg/mL of silica coated gold nanorods,the viability of cells in the medium with or without serum was 99.0% and 85.1%,respectively. However,the viability of cells cultured with or without serum decreased to 58.3% and 31.2%,respectively when the concentration of silica coated gold nanorods was 1 250 μg/mL. The serum in the culture medium prevented HeLa cells from taking up the silica coated gold nanorods,and the uptake was time-dependent. The internalized silica coated gold nanorods mainly accumulated in the lysosomes,did not enter the mitochondria. The study could provide a reference for the imaging and treatment of cervical cancer with gold nanorods in the future.
		2020 Vol. 39 (4): 466-472 [Abstract] ( 180 ) HTML (1 KB) PDF (18795 KB) ( 43 )

Reviews

	473	Review on the Application of Deep Learningin Video Analysis of Minimally Invasive Surgery
		Shi Pan, Zhao Zijian
		DOI: 10.3969/j.issn.0258-8021.2020.04.011
		Deep learning theory has been widely used in the video analysis of minimally invasive surgery,and has made remarkable achievements in the surgical tool detection and tracking,surgical tool presence detection and workflow recognition of minimally invasive surgery. In the long run,the detailed analysis of the video content of minimally invasive surgery can not only automatically identify the ongoing surgical tasks,but also be used to remind clinicians of possible complications. In recent years,with the continuous development of technology,the application of deep learning in the video analysis of minimally invasive surgery has made great progress. This paper firstly expounded the significance,difficulties and the relevant technical content of video analysis of minimally invasive surgery,mainly introducing the advantages based on deep learning algorithm. This paper also summarized the research achievements of deep learning in the field of surgical tool detection and tracking,surgical tool presence detection and workflow recognition of minimally invasive surgery,classified and concluded algorithms based on their features in different fields of minimally invasive surgery video analysis,and evaluated their results. In the end,this paper summarized and looked forward to the future development of minimally invasive surgery video analysis.
		2020 Vol. 39 (4): 473-484 [Abstract] ( 373 ) HTML (1 KB) PDF (1113 KB) ( 380 )

	485	A Review of MRI-Based Synthetic CT Image Generation
		Jian Yingchao, Fu Dongshan, Wang Wei
		DOI: 10.3969/j.issn.0258-8021.2020.04.012
		Radiation therapy is one of the important ways in tumor treatments,and CT is the main reference image of current radiation therapy. Compared with CT,MRI has good soft tissue contrast and is completely harmless to the human body. In recent years,MRI has been increasingly used for soft tissue delineation and radiation therapy guidance. However,the ionizing radiation of CT scan affects the health of patients,at the same time,scanning CT and MRI increase the economic burden of patients,and the registration fusion of CT and MRI introduces systematic errors.Radiotherapy using MRI only has received more and more attention from researchers. Nevertheless,MRI has no connection with electron density and cannot be directly used for dose calculation and X-ray based patient placement verification,therefore it is necessary to study the correlation algorithm to obtain the electron density information or HU value of the patient tissue according to the MRI image,that is to generate synthetic CT or pseudo CT. Currently existing methods for generating synthetic CT or pseudo CT are classified into three categories,including voxel-based,atlas-based and hybrid methods. According to the applied methods,we summarized and analyzed the sequence images,the number of applications and the anatomical parts. This study may represent a new direction in radiation therapy,which can avoid the use of ionizing radiation from traditional CT,and can use high-resolution MRI images to monitor tumors and enable patients to perform more precise radiation therapy.
		2020 Vol. 39 (4): 485-492 [Abstract] ( 487 ) HTML (1 KB) PDF (877 KB) ( 546 )

	493	Research Progress of Biomimetic Periosteum for Bone Tissue Regeneration
		Liu Laijun, Zhang Yu, Li Chaojing, Mao Jifu, Wang Fujun, Wang Lu
		DOI: 10.3969/j.issn.0258-8021.2020.04.013
		Periosteum is a highly vascularized connective tissue membrane covering the outer surface of cortical bone (except joints).It contains various bone cells and growth factors and plays a vital role in the development and regeneration of bone tissue. However,due to the limited number of available healthy periosteum,clinical treatment of critical size bone defects caused by trauma,tumors and congenital diseases still faces significant challenges. This has provided the impetus for the development of periosteum substitutes (biomimetic periosteum),which have a structure and function similar to natural periosteum. In this review,we highlighted the current research status in properties of biomimetic periosteum at the macro and micro levels,including biocompatibility,graded functionality,mechanical stability,biological activity,and clinical manageability. The four biomimetic periosteum preparation methods were listed based on different principles,and the main defects of various preparation methods were discussed. It is expected to provide reference for the development of biomimetic periosteum with better performance.
		2020 Vol. 39 (4): 493-503 [Abstract] ( 410 ) HTML (1 KB) PDF (8327 KB) ( 187 )

Communications

	504	Study on the Clinical Effect of Multi-Beam UltrasoundTreatment for Breast Cancer
		Qi Xiaodong, Liu Xinghua, Feng Lihui, Liang Shuang, Yang Zibin, Ni Guangnan, Zeng Ziheng
		DOI: 10.3969/j.issn.0258-8021.2020.04.014

		2020 Vol. 39 (4): 504-507 [Abstract] ( 302 ) HTML (1 KB) PDF (8847 KB) ( 129 )

	508	Effect of Low-Frequency Healthy Side rTMS with High Dose on Upper Limb Motor Function in Cerebral Infarction Patients
		Wang Yuqin, Lv Mingxin, Liu Shuangjie, Liang Junjun, Li Tingting
		DOI: 10.3969/j.issn.0258-8021.2020.04.015

		2020 Vol. 39 (4): 508-512 [Abstract] ( 230 ) HTML (1 KB) PDF (692 KB) ( 142 )

Copyright © Editorial Board of Chinese Journal of Biomedical Engineering
Supported by:Beijing Magtech