Chinese Journal of Biomedical Engineering

Home

Editorial Board

Download

Quick Search

Office Online

Current Issue

Download Articles

Download

Links

2018 Vol. 37, No. 1 Published: 2018-01-20

Reviews Communications Regular Papers

Regular Papers

	1	Diagnosis of Alzheimer′s Disease via Multi-Modal Canonical Feature Representation
		Zhuo Yinan,Yang Peng,Deng Yun, Ni Dong,Lei Baiying,Wang Tianfu
		DOI: 10.3969/j.issn.0258-8021.2018.01.001
		To better explore the underlying disease patterns, a study of modeling the disease procession of Alzheimer′s disease (AD) for early diagnosis of mild cognitive impairment(MCI)is necessary. In this work, we proposed a new method to represent the multi-modal correlations of ROI features for AD diagnosis. First, we applied canonical correlation analysis (CCA) to discover the relationships of ROIs among different modalities and specifically with sparse least square regression loss function to acquire the discriminative features. Then we trained a classification model via support vector machine (SVM) by using the selected features for diagnosis. The empirical studies with 805 subjects downloaded from the Alzheimer′s Disease Neuroimaging Initiative (ADNI) database showed that our method achieved promising improvement when compared with others methods. We achieved 92.01% in AD vs NC (normal control), 74.83% in MCI vs NC and 70.27% in p-MCI (progressive-mild cognitive impairment) vs s-MCI (stable-mild cognitive impairment). In conclusion, the proposed method is beneficial to the early diagnosis of the disease.
		2018 Vol. 37 (1): 1-7 [Abstract] ( 461 ) HTML (1 KB) PDF (868 KB) ( 924 )

	8	Multichannel EEG-EMG Coupling Analysis Using aVariable Scale Symbolic Transfer Entropy Approach
		Gao Yunyuan, Ren Leilei, Zhou Xu ,Zhang Qizhong, Zhang Yingchun
		DOI: 10.3969/j.issn.0258-8021.2018.01.002
		The coupling strength (CS) of the electroencephalogram (EEG) and electromyogram (EMG) reflects the connection between the cerebral cortex and active muscles in motion control. Conventional symbolic transfer entropy analysis (STE) methods are limited to effectively reveal the dynamic characteristics. In order to quantitatively analyze multi-channel EEG and EMG coupling, a variable scale symbolic transfer entropy (VSP-STE) analysis approach was proposed in this paper, and a CS expressive method was presented to quantitatively measure the corticomuscular CS of EEG and EMG signals during grip tasks. The influence of the scale parameter on EEG-EMG transfer entropy was first analyzed and compared at different grip force levels. The optimized scale parameter was then applied to the STE calculation in experimental data analysis. Results demonstrated the dominance of the C3 and C4 EEG channels in motion control, as well as the contralateral control mechanism of the brain. In addition, the premotor cortex (FC5 and FC6) also played an important role in corticomuscular coupling. Results showed that the EMG→EEG transfer entropy increased as the grip force increased (5, 10, 20 kg), and the right hand (dominant hand) showed higher EEG→EMG transfer entropy than the left hand in all the 3 subjects. The two-way (EEG→EMG, EMG→EEG) coupling strengths were also increased as the grip force increased. The statistical analysis showed that at 5, 10, 20 kg grip force levels, EMG→EEG coupling strength of left hand was 0.0330±0.0058,0.0373±0.0040, 0.0451±0.0055 respectively, and that of right hand was 0.0352±0.0029, 0.0432±0.0035, 0.0603±0.0018 respectively. There was a significant difference (p<0.05) of coupling strength, except for that between 5 kg and 10 kg of grip force in left hand. For EEG→EMG, coupling strength of left hand is 0.0253±0.0047, 0.0379±0.0026, 0.0481±0.0068, and that of right hand is 0.0333±0.0041, 0.0510±0.0057, and 0.0649±0.0085 respectively. The difference of coupling strength was significant for all. In conclusion, our results show that corticomuscular coupling is a two-way process and the coupling strength varies with different channels and gripping strength. VSP-STE can be used to describe the nonlinear synchronizing characteristics and information interaction of the cerebral cortex and neuromuscular tissue.
		2018 Vol. 37 (1): 8-16 [Abstract] ( 561 ) HTML (1 KB) PDF (935 KB) ( 815 )

	17	A Peripheral Blood WBC Classification with ConvolutionalNeural Network
		Chen Chang, Cheng Shaojie, Li Weibin,Chen Min
		DOI: 10.3969/j.issn.0258-8021.2018.01.003
		The automatic classification of white blood cell (WBC) image is essential because it helps to enhance the efficiency of clinical diagnosis and treatment. However, the classification accuracy is still need to be boost for adapting to practical applications. In this paper, we proposed an automatic classification method based on the convolution neural network (CNN). We tentatively fed our training dataset into AlexNet and LeNet using a widely used deep learning platform Caffe. Five classes of WBCs images collected by a CellaVision DM96 in peripheral blood smears were adopted as the training dataset. These manual labeled images were apportioned into three groups (training, validation and testing) randomly to construct the original dataset according to the proportion of 7：2∶1. With the augmentation methods, such as rotation and mirror, we expanded the original dataset. Stochastic gradient descent algorithm was adopted as the optimizing method for training CNNs. The experimental results demonstrated that the network structure of AlexNet was unsuitable to achieve the ideal classification accuracy which more than 95%. While the network structure of LeNet had achieved the expected target. However, the more massive and more time consuming of LeNet suggested us to further optimize the connection of layers to derive a new network with lightweight structure, named as CCNet. The model size, time for training, and time for evaluation of CCNet were only 1/1000, 1/3, and 1/30 compared with LeNet, respectively. The best classification accuracy of CCNet and LeNet for five classification of WBCs was 99.69% and 99.18% with 979 WBC images, higher than those of the previous reports. It demonstrated that CNNs especially CCNet had clear advantages than previous works both in classification accuracy and speed.
		2018 Vol. 37 (1): 17-24 [Abstract] ( 577 ) HTML (1 KB) PDF (872 KB) ( 1101 )

	25	Parameter Optimization in Face-Based P300 Speller System
		Sun Hongyan, Jin Jing, Zhang Yu, Wang Bei ,Wang Xingyu
		DOI: 10.3969/j.issn.0258-8021.2018.01.004
		The P300 based BCI is often used in speller system, because of its high accuracy and information transfer rate. Previous studies showed that face stimulus could induce recognizable event related potentials, which improve the performance of a P300 speller system. However, the form of face stimulus presentation also directly affects the system′s performance. The experiments we executed were under three different parameters: stimulus onset asynchrony (SOA) (130ms vs 200ms), the size of screen (15.6inch vs 24inch) and image pixels (50×69 vs 80×110), and 10 healthy subjects were invited to participate in this experiment. The results showed that enlarging the three parameters all improved the per-trial classification accuracy during the offline training. However, in the process of online testing, only the paradigm with 200 ms stimulus intervals achieved significantly higher online classification accuracy than that with 130 ms stimulus intervals (90%±7% vs 75%±13%). Besides, the three paradigms’ parameter adjustments have different influences on the amplitude of ERP, such as N200, P300, N400, et al. In the practical usage of P300 speller system, parameter optimization should be taken into consideration to improve the system′s performance.
		2018 Vol. 37 (1): 25-32 [Abstract] ( 254 ) HTML (1 KB) PDF (841 KB) ( 642 )

	33	Research on EEG Phase Amplitude Modulation during Epileptic Seizure
		Cao Chunyu， Han Ling， Li Chunsheng
		DOI: 10.3969/j.issn.0258-8021.2018.01.005
		EEG is the most important tool in the diagnosis and treatment of epilepsy, and the analysis of large data volume of EEG records has brought considerable difficulties, which may be overcome by the automatic computer assistant classification tools. In this paper, the coupling between the phase of low frequency and the amplitude of high frequency was studied from the perspective of phase amplitude modulation, and the normalized modulation index (MI) was further used to quantify the coupling intensity in different frequency pairs. The Bonn datasets with total 200 samples of interictal and ictal EEG segment were used in this study. A method of partitioning the MI map according to the ranges of high and low frequency rhythms was proposed. Results have shown that the MI value between Gamma and Delta (2～4 Hz) rhythm was significantly increased (P<0.01) from the interictal state (0.0036±0.0087) to the ictal state (0.0099±0.0096). The MI value of Theta (4～8 Hz) Gamma was significantly increased (P<0.01) from the interictal state (0.0014±0.0032) to ictal state (0.0087±0.0062). Modulation strength of Theta Beta rhythm also increased significantly from interictal state (0.0005±0.0007) to ictal state (0.0022±0.0013). The classification accuracy based on MI characteristics was 97% for the distinguishing the ictal data, and the accuracy was not changed much by using support vector machine with five folder cross validation and random forest. The using of proposed method would greatly improve the efficiency of clinical video EEG analysis for determining the ictal state.
		2018 Vol. 37 (1): 33-39 [Abstract] ( 374 ) HTML (1 KB) PDF (840 KB) ( 632 )

	40	Quantitative Measurement of 3D Skin Wound Area Based on Single Camera
		Liu Chunhui, Fan Yubo, Xu Yan
		DOI: 10.3969/j.issn.0258-8021.2018.01.006
		The measurement of skin wounds is an important work in the field of clinical research and forensic identification. Quantitative measurement of skin wounds area has important significance in forensic identification, clinical trials, the wound pathological analysis and daily patient care. In this paper, the algorithm of structure from motion (SFM) and least squares conformal mapping (LSCM) were introduced into the measurement of skin wounds area, combined with image segmentation, an algorithm that suitable for the area measurement of the human body surface was proposed. This paper used eigenvalue extraction, sparse reconstruction and dense reconstruction in order to get the 3D point cloud of the tested body. Then the paper used the Poisson reconstruction algorithm on the point to make them networked, and unwrap the UV map of the 3D model for skin wounds extraction and measurement in the end. This paper used the known area simulation wound as a benchmark, 40 groups were adopted to evaluate the accuracy of the algorithm. The measuring accuracy of the experimental results showed that the algorithm in this paper reached the accuracy of 0.97, compared with 2D measurement method, the accuracy was increased by 10.79%. The algorithm in this paper solved the shortcoming of the contact method and the problem of human curvature or angle deflection, which was hard to solve by the 2D measurement. In addition, the algorithm in this paper has less dependence of equipment and high accuracy especially in the parts with large curvature.
		2018 Vol. 37 (1): 40-48 [Abstract] ( 301 ) HTML (1 KB) PDF (791 KB) ( 389 )

	49	The Design and Test of Pulsating Blood Pump Driven by Direct-Current Electromagnetism for Temporary Assistance
		Liu Jingjing, Ge Bin, Lu Tong ,Wei Lingxuan ,Zhang Lei ,Dong Jiaoyang
		DOI: 10.3969/j.issn.0258-8021.2018.01.007
		The aims of this work are to design a pulsating blood pump driven by direct-current electromagnetism and to evaluate the performance indices of prototype. First of all, the pulsating blood pump driven by direct-current electromagnetism was designed with both the drive method of the reciprocating rectilinear motion of a magnet implemented by utilizing solenoids with direct current and the structure of compensation solenoid within uniform magnetic field which was designed through numerical simulation. Next, the value of magnetic driving force created by the blood pump was measured with the prototype and the acceleration test bed when the different current was applied. After that, none heating problem of energized solenoids was detected through calculation and experiment. Finally, the flow performance index of prototype was measured in the conditions that the ranges of preload or afterload were 5-30 mmHg and 50-80 mmHg respectively through the flow test bed. The magnetic driving force created by the prototype possessed the positive linear relation to the current and accorded with the design requirement when the current was 2.7 A. When the current was 2.7 A and the driving frequency was 80 per min, the temperature on the internal surface of energized solenoids contacting with the blood was raised 1 ℃ and stabilized at 27 ℃, the flow of prototype was over 3.0 L·min^-1 except that the gap between preload and afterload was more than 70 mmHg. The pulsating blood pump driven by direct-current electromagnetism reached the clinical requirement of the perfusion of organs isolated and the temporary assistance of extracorporeal circulation, and the design was of significance to the development of extracorporeal circulation blood pump.
		2018 Vol. 37 (1): 49-56 [Abstract] ( 263 ) HTML (1 KB) PDF (799 KB) ( 514 )

	57	*Characterization of in vivo* Bioelectronic Nose with Combined #br# Manganese-Enhanced MRI and Brain-Computer Interface**
		Zhang Bin,Wu Zhe, He Hongjian,Ding Qiuping,Qin Zhen,Gao Keqiang,Zhong Jianhui, Wang Ping
		DOI: 10.3969/j.issn.0258-8021.2018.01.008
		By using mammal′s own olfactory system, sensitivity and specificity of the in vivo bioelectronic nose is substantially enhanced. However, the specific region in the olfactory bulb where to implant the electrode has been based on the researcher′s experiences, which often results in unsatisfied success rate. This study takes advantage of the paramagnetism and calcium ion similarity of the manganese ion. A small dose of manganese ion was delivered into the right naris of 10 rats, an odor was delivered to the nose of the rat, and a series of magnetic resonance images (MRI) were taken. With the MRI data, a region in the olfactory bulb activated by the specific odor was identified. Micro-wire electrode was implanted into the region and olfactory signals were recorded. When the rat was stimulated by the specific odor, the β wave of the LFP was found to increase and the spike signals also had responses. Linearity was observed between the difference of the spike-firing rate caused by the odor stimulation and the concentration of the odor. The limits of detection to isoamyl acetate and n-butyric acid were determined to be 0.033μM and 0.0072μM, respectively. As the first bioelectronic nose assisted by manganese-enhanced MRI, it has a promising future in explosives searching or food safety.
		2018 Vol. 37 (1): 57-63 [Abstract] ( 351 ) HTML (1 KB) PDF (796 KB) ( 360 )

	64	Mass Transport Modeling on Cryoprotectant Permeation into Articular Cartilage
		Yu Xiaoyi, Zhang Shaozhi ,Chen Guangming ,Li Wei
		DOI: 10.3969/j.issn.0258-8021.2018.01.009
		The knowledge of permeation kinetics of cryoprotective agent (CPA) in articular cartilage (AC) is of critical importance for designing optimal CPA addition protocols to achieve successful cryopreservation of AC. In this study, a mass transport model was developed for predicting CPA diffusion in AC. The effective diffusion coefficient was correlated to the infinite dilution coefficients through a binary diffusion thermodynamic model. The UNIFAC model was used to evaluate the activity coefficient, the Vignes equation was employed to estimate the composition dependence of the diffusion coefficient, and the Siddiqi-Lucas correlation was applied to determine the diffusion coefficient at infinite dilution. The model was demonstrated to predict the spatial and temporal CPA distribution in AC during addition. Results showed that the predicted overall CPA uptakes by AC match the experiment data well. The mean relative error (MRE) and coefficient of determination (R²) are 1.90%～36.29% and 0.959～0.998 for dimethyl sulfoxide, 13.56%～19.19% and 0.990～0.995 for glycerol, 8.89%～22.09% and 0.969～0.988 for ethylene glycol, and 5.35%～23.76% and 0.971～0.992 for propylene glycol, respectively.
		2018 Vol. 37 (1): 64-71 [Abstract] ( 236 ) HTML (1 KB) PDF (821 KB) ( 263 )

	72	Binocular Stereo Vision Based Real-Time Trackingfor Respiratory Motion
		Wang Yan, Sun Xiangming, Xiong Poyi ,Wang Yufei
		DOI: 10.3969/j.issn.0258-8021.2018.01.010
		In radiotherapy, in order to reduce the impact of respiration motion on precise radiotherapy, we developed a binocular stereo vision based real-time tracking for respiration motion to reduce the treatment error due to the tissues moving in the tumor target. During radiotherapy, the binocular above the treatment bed were used to collect the images of the markers and transmit them to the computer. The cosine algorithm identified the markers that present on the thoraco-abdominal surface. The parallax of binocular matched the images. Basic principles of binocular stereo vision and aperture imaging to calculate the three-dimensional coordinates of the markers. By monitoring the three-dimensional coordinates of the markers, to determine whether the markers are moving. In experiment, the three-dimensional coordinates of nine markers were measured by real-time, experimental results showed that the error between the measured coordinate value of the surface markers and the actual coordinate average value was less than ± 1 mm, the rmse was less than 0.12 mm, it took 35 ms to calculate the three-dimensional coordinates of the nine markers. The tumor motion tracking method based on binocular stereo vision is a high precision, real-time and stability method, reducing the impact of respiration motion on precise radiotherapy.
		2018 Vol. 37 (1): 72-78 [Abstract] ( 355 ) HTML (1 KB) PDF (752 KB) ( 328 )

	79	Effect of Electrospun Collagen/Chitosan Composite Nanofibrous Membrane on Osteogenesis for Bone Regeneration
		Li Xiaojing ,Gao Bo, Dong Yan ,Gou Zhongru, Miao Yuwen
		DOI: 10.3969/j.issn.0258-8021.2018.01.011
		The aims of this study are to prepare collagen/chitosan composite nanofibrous membranes, and to examine their biocompatibility and effect of osteogenic differentiation for bone marrow mesenchymal stem cells(BMSCs). Take acetic acid as solvent,poly(ethylene oxide) (PEO) as plasticizer, the collagen nanofibrous membranes (the mass rate of collagen/PEO, 4∶1) and various collagen/chitosan nanofibrous membranes (the mass rate of collagen/chitosan/PEO, 5∶1∶4,5∶2∶3,5∶4∶1) were fabricated through electrospinning technique. The surface morphology of membranes were observed by scanning electron microscopy (SEM). BMSCs were cultured on the collagen nanofibrous membranes and the optimized collagen/chitosan nanofibrous membranes, the cell proliferation and osteogenic makers including alkaline phosphate(ALP), collagen content, osteocalcin (OCN) and mineralized matrix deposits were detected respectively. SEM showed collagen nanofibers and collagen/chitosan nanofibers of mass ratio of 5∶1∶4 had a mean fiber diameter. Both collagen nanofibrous membranes and collagen/chitosan nanofibrous membranes supported adhesion, proliferation and osteogenic differentiation of the BMSCs. The collagen/chitosan nanofibrous membranes induced higher proliferation of BMSCs than collagen nanofibrous membranes. The collagen contents were higher in cells cultured on the collagen/chitosan nanofibrous membranes for 14 days as compared to collagen nanofibrous membranes (p<0.05). Similarly, cells cultured on collagen/chitosan nanofibrous membranes had the higher expression of ALP, OCN and mineralized matrix deposits than that on collagen nanofibrous membranes. Our results demonstrated that the collagen/chitosan composite nanofibrous membranes promoted adhesion, proliferation and osteogenic differentiation of BMSCs,implying their promising applications for bone regeneration.
		2018 Vol. 37 (1): 79-85 [Abstract] ( 331 ) HTML (1 KB) PDF (902 KB) ( 259 )

Reviews

	86	Research Progress of Medical Image Recognition Based on Deep Learning
		Liu Fei ,Zhang Junran ,Yang Hao
		DOI: 10.3969/j.issn.0258-8021.2018.01.012
		In recent years, with the rapid development of medical imaging technology, medical image analysis has entered the era of big data. How to extract useful information from a large number of medical image data has become one great challenge to medical image recognition. Deep learning is a new field of machine learning, conventional machine learning method can’t effectively extract enough information contained in the medical image, while the deep learning has the power of establishing a hierarchical model, powerful automatic feature extraction, complex model building and efficient feature expression through the simulation of the human brain. More importantly, deep learning method can extract the features from the bottom to the top level from the original data of the pixel level, which provides a new way to solve the new problems faced by medical image recognition. Based on a large number of domestic and foreign literatures, this paper elaborated the three methods of depth learning, enumerated three common implementation models of deep learning methods, and introduced the training process of depth learning. We summarized the application of deep learning in two aspects of disease detection and classification and lesions recognition, and summarized the two common problems in the application of deep learning in medical image recognition. The analysis and prospects of deep learning in medical image recognition problems were proposed and discussed as well.
		2018 Vol. 37 (1): 86-94 [Abstract] ( 915 ) HTML (1 KB) PDF (748 KB) ( 2130 )

	95	Deep Learning in Digital Pathology Analysis
		Yan Wen, Tang Ye ,Chang Eric I-Chao, Lai Maode ,Xu Yan
		DOI: 10.3969/j.issn.0258-8021.2018.01.013
		Pathology is regarded as the gold standard for diagnosis of cancer. Pathological analysis and prognosis are usually performed by pathologists, however, it could be time-consuming and labor-intensive. As the development of the whole slide pathology, it is thanks to artificial intelligence (AI)gradually promotethe transition from qualitative analysis to quantitative analysis. In recent years, the AI technology, especially deep neural network, has greatly promoted the progress of pathological diagnosis, which is more intelligentialized, accurate and repeatable. This paper describes the basic concept of deep learning and its application in digital pathology analysis. We give a brief overview of the application of deep learning in detection and segmentation of cell and tissue, classification and grading of cancer, and other applications. Finally, we propose the existing problems and the prospect of future development in the analysis of digital pathology.
		2018 Vol. 37 (1): 95-105 [Abstract] ( 414 ) HTML (1 KB) PDF (912 KB) ( 999 )

	106	Research Advancements of Transcranial Electrical Stimulation on Motor Function Recovery after Stroke
		Mu Siyu, Xu Minpeng, He Feng ,Zhang Lixin,Ming Dong
		DOI: 10.3969/j.issn.0258-8021.2018.01.014
		Stroke is one of the main causes for motor dysfunction, which brings about great spiritual and economic burdens for the society and families. Motor rehabilitation therapies using transcranial electrical stimulation (TES) provide a promising treatment for patients to improve their motor functions and life qualities. TES is a painless, noninvasive brain stimulation method, which has showed the effects of regulating calcium concentration in neurons, enhancing synaptic plasticity, modulating neural firing frequency and changing cortical excitability, which in turn improved the neuronal function. This paper reviewed the application of TES to motor function recovery after stroke, including neural effects, selection of parameters, safety evaluation and achievements as well as present issues for further investigations in both scientific research and clinical tests.
		2018 Vol. 37 (1): 106-111 [Abstract] ( 293 ) HTML (1 KB) PDF (705 KB) ( 981 )

	112	Research Progress of the Methods Improving Porosity and Cellular Infiltration in Articular Cartilage Extracellular Matrix Scaffold
		Zhao Zhongqi ,Wang Xiaokun ,Peng Huimin ,Liang Jiadi ,Wang Xiaojin,Liu Chang
		DOI: 10.3969/j.issn.0258-8021.2018.01.015
		The regeneration and repair of articular cartilage injury has been a clinical challenge. There has been an increasing focus on developing cartilage tissue engineering due to the limitations of surgical treatment. In cartilage tissue engineering, cartilage matrix is a promising biomaterial for cartilage regeneration given the evidence supporting its chondroinductive character. This natural biomaterial serves as a scaffold which can provide collagen network and biological factors for cartilage tissue regeneration. Decellularization was optimized to remove cells and cell remnants from cartilage effectively. However, the reduced vascularity, limited cell population, and dense extracellular matrix (ECM) inhibit cartilage regeneration. Therefore, most cartilage tissue culture studies focus on improving porosity of the scaffold and promoting cellular migration into central regions. This review focused on recent studies on cartilage ECM scaffold and briefly introduced key ways to obtain high-porosity scaffold that can improve cell infiltration and migration. We summarized from several aspects including decellularization, concentration regulation, oriented freeze-drying, artificial channels and laser micropatterning. By analyzing the principles and advantages of different methods, the importance of pore structure is further emphasized, and the relationship between scaffold performance and regulation methods is clarified.
		2018 Vol. 37 (1): 112-118 [Abstract] ( 310 ) HTML (1 KB) PDF (704 KB) ( 524 )

Communications

	119	3D Topographic Reconstruction of Optic Nerve Head Based on SFS Approach
		Wang Peipei, Sun Jiuai
		DOI: 10.3969/j.issn.0258-8021.2018.01.016

		2018 Vol. 37 (1): 119-123 [Abstract] ( 254 ) HTML (1 KB) PDF (781 KB) ( 284 )

	124	Establishment of New Bone Tissue Engineering Three-Dimensional Complex in vitro
		Maimaitiaili Abulikemu, Wang Tengfei, Tao Ying ,Wang Xiaoshuai ,Xu Leilei ,Song Xinghua
		DOI: 10.3969/j.issn.0258-8021.2018.01.017

		2018 Vol. 37 (1): 124-128 [Abstract] ( 262 ) HTML (1 KB) PDF (858 KB) ( 229 )

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