Chinese Journal of Biomedical Engineering

Home

Editorial Board

Download

Quick Search

Office Online

Current Issue

Download Articles

Download

Links

2018 Vol. 37, No. 3 Published: 2018-07-20

Reviews Communications Regular Papers

Regular Papers

	257	The Color Fundus Image Enhancement Algorithm Based on Retinex Theory
		Liu Yuhong, Yan Hongmei
		DOI: 10.3969/j.issn.0258-8021.2018.03.001
		The color fundus image is usually suffered from poor brightness, low contrast and local detail loss. This paper analyzed drawbacks of Retinex methods, and proposed a new effective enhancement algorithm based on Retinex theory. First of all, luminance components of original image were extracted. Next, a multi-scale retinex algorithm was used on it. The simplest possible color balance algorithm was adopted to modify the gain/offset correction method.At last, we calibrated the red channel information to restore the luminance information. In order to verify the effectiveness of the method, the proposed method was compared with other enhancement algorithms including multi-scale retinex(MSR), multi-scale retinex with color restoration (MSRCR), histogram equalization(HE), contrast limited adaptive histogram equalization (CLAHE) on the DIARETDB0 fundus image database. Experimental results showed that the proposed method had better effect on the color protection, vascular contrast improvement and enhance image details than the other Retinex algorithms and conventional image enhancement methods. The information entropy was increased by 5% to 7% and the peak signal-to-noise ratio (PSNR) was 1～2 times higher than the conventional methods. The objective image quality index was significantly better thanthe other fundus image enhancement methods. This method is of significance to further fundus image recognition.
		2018 Vol. 37 (3): 257-265 [Abstract] ( 697 ) HTML (1 KB) PDF (10807 KB) ( 302 )

	266	Brain Extraction Method Based on SIFT Feature Matching among Key Points
		Jiang Shaofeng, Huang Zhipeng, Yang Suhua, Chen Zhen, Zhang Congxuan
		DOI: 10.3969/j.issn.0258-8021.2018.03.002
		Brain extraction is an important preprocessing step in most of image analyses for cerebral MRI image. ASIFT(Scale Invariant Featun Transform) feature matching based brain extraction method was proposed in this paper for the precise and robust brain extraction. This method combined the tessellated surface based method (BET) and atlas based method. Firstly,the BET method was used to evolve the brain contour, and the SIFT features of the key points on brain contour were extracted iteratively to find the matched key points between the target image and atlas image using weak constraint of spatial distance based matching method. Next, the matched points were used to renew the parameter of BET and the brain contour iteratively, thus a robust brain contour was obtained. At last, a graph cuts method was used to refine the brain contour to obtain a precise brain contour. Twenty MRI volumes from IBSR web site were tested using this method and other three methods, and the proposed method obtained the highest Dice precision (0.962±0.008), Jaccard precision (0.926±0.014) and specificity (0.994±0.004), as well as the lowest false positives rate (4.95%±2.74%) and lower false negatives rate (2.82%±2.0%). The highest extraction precisions demonstrated this method precise, and the lower corresponding standard deviations showed this method was robust. Therefore, this proposed method is a precise and robust brain extraction method.
		2018 Vol. 37 (3): 266-273 [Abstract] ( 487 ) HTML (1 KB) PDF (7568 KB) ( 352 )

	274	Melanoma Recognition in Dermoscopy Images via Deep Residual Network
		Li Hang, Yu Zhen, Ni Dong, Lei Baiying, Wang Tianfu
		DOI: 10.3969/j.issn.0258-8021.2018.03.003
		Malignant melanoma is one of the most common and deadly skin cancers. Clinically, dermoscopy is a routine method for early diagnosis of malignant melanoma. However, human's visual examinations are laborious, time-consuming, and highly dependent on dermatologist’s clinical experience. Therefore, it is important to design an algorithm for recognizing melanoma automatically in dermoscopy images. This study proposed a novel framework for the evaluation of dermoscopy images, using deep learning to generate more discriminative features with limited training data. Specifically, we first extracted the intermediate convolutional features of each skin lesion image using a very deep residual neural network including 152 network layers (i.e. Res-152) which was pre-trained on a large natural image dataset, and the final deep representation was obtained by averaging the spatial feature maps into single feature vector, then, the support vector machine (SVM) was used to classify the melanoma. By using the proposed method 248 melanoma images and 1031 non-melanoma images in published ISBI 2016 challenge datasets of skin lesion images were evaluated, obtaining accuracy rate of 84.96% and AUC of 84.18%. In addition, in order to demonstrate the effect of neural network depth on the classification results, we compared the different depth of the model framework. Our approach, which could solve large variations in melanoma and small differences between melanoma and non-melanoma with the limited training data, can produce more discriminative representations than existing studies using hand-crafted features (i.e. the BoF models based on densely sampled SIFT (DSIFT) descriptors) or only to extract features from the fully connected layers.
		2018 Vol. 37 (3): 274-282 [Abstract] ( 806 ) HTML (1 KB) PDF (7496 KB) ( 886 )

	283	Objective Discrimination of Depression: Detection and Analysis of Resting State Functional Connectivity Based on Optical Brain Imaging
		Zhu Huilin, Xu Jie, Li Jiangxue, Peng Hongjun
		DOI: 10.3969/j.issn.0258-8021.2018.03.004
		Recently, resting-state functional connectivity (RSFC) has gradually been studied in patients with mental disorders by functional near-infrared spectroscopy (fNIRS). However, it is still unknown whether RSFC derived from fNIRS is predictable for depressive disorders. In this work, we employed fNIRS(42 channels) to measure 8-minute spontaneous hemodynamic activity in the prefrontal cortex (PFC) of 28 patients having depressive disorders and 30 healthy controls. After filtering irrelative components by independent component and band-pass filter (0.008-0.09 Hz), we calculated left-right correlations in the prefrontal cortex which included inferior prefrontal cortex (IFG), middle prefrontal cortex (MFG) and superior prefrontal cortex (SFG).Then we selected two significant parameters (left-right correlations in the IFG and MFG as a participant’s two features for further classification (75% of the participants) and prediction (25% of the participants) using linear discriminant analysis (LDA) and support vector machine (SVM). Finally, a sensitivity of 73-74% and specificity of 83-87%was yielded. These results supported that RSFC derived from fNIRS is a feasible and effective technique to identify whether someone is suffered from depressive disorders.
		2018 Vol. 37 (3): 283-289 [Abstract] ( 444 ) HTML (1 KB) PDF (4458 KB) ( 510 )

	290	The Effects of rTMS Combined with Motor Training on Brain Network in Resting Status
		Jin Jingna, Wang Xin, Lin Yu, Zhang Kai, Li Ying, Xiang Fang, Liu Zhipeng, Yang Xuejun, Yin Tao
		DOI: 10.3969/j.issn.0258-8021.2018.03.005
		It was reported that repetitive transcranial magnetic stimulation (rTMS) combined with motor training could improve the motor skill, which could be used in motor rehabilitation after stroke. In this study, the effects of rTMS combined with motor training on brain neural activities were investigated based on the method of brain network. Ten healthy volunteers were recruited. The 1 Hz rTMS over the dominant hemisphere combined with unfamiliar motor training with non-dominant hand subsequently was performed rTMS combined with motor training lasted 14 days to improve the motor function of non-dominant hand. Electroencephalography (EEG) in resting state with eyes closed was recorded before and after rTMS combined with motor training. The functional connectivity was analyzed using the method of phase lag index (PLI). We constructed weighted network and calculated the network topology characteristics based on PLI, subsequently. Finally, the signed-rank test was used for statistical analysis. We found that the changes of functional connectivity could be detected mainly between functional regions rather than inside regions. The functional connectivity at lower frequency band (theta and alpha) was significantly increased, and was opposite at higher frequency band (beta, gamma₁ and gamma₂). Furthermore, we found that the rTMS combined with motor training had a significant influence on the functional connectivity between central region in non-dominant hemisphere and dominant frontal regions (before: 0.141 4±0.102 5;after:0.217 2±0.134 7; P<0.05) and non-dominant frontal regions(before:0.141 0±0.109 9;after:0.205 9±0.136 1; P <0.05) at alpha frequency. Node efficiency increased at low band and decreased at high band, and node path length was opposite. Specifically, the node efficiency at gamma₂ bandchanged significantly, mainly in central regions of both hemisphere (left, before: 0.060 0±0.000 3; after: 0.042 9±0.001 3; P <0.05; right, before: 0.060 7±0.002 3; after: 0.041 9±0.002 4; P <0.05), and also the node path length (left, before:18.539 0±0.457 1;after:28.585 8±1.001 4;P <0.05; right, before: 18.650 8±0.438 6; after: 28.853 0±1.652 6;P <0.05). This study was helpful to understand the brain mechanism of rTMS combined with motor training on improvement of motor skill, and comprehend the impact of stroke and brain lesions on brain activities.
		2018 Vol. 37 (3): 290-296 [Abstract] ( 580 ) HTML (1 KB) PDF (3017 KB) ( 603 )

	297	Effects of Emotional State on Working Memory Updating Based on ERP
		Tong Jingjing, Liu Shuang, Guo Dongyue, Ke Yufeng, Ming Dong
		DOI: 10.3969/j.issn.0258-8021.2018.03.006
		Researcher shave payed increasing attention to studies of the emotional effect on working memory, especially on the updating function of central executive system. In this paper, we discussed the effects of positive, neutral and negative emotions on updating of working memory. Sixteen subjects participated in this experiment, in which emotional pictures were firstly presented to evoke their three emotional states followed by 4 numbers running memory task. Results showed that the updating time in the positive state is significantly longer than that in the neutral state (Positive state: (1.016±0.338)s; Neutral state:(0.814±0.347)s, P<0.05), and the amplitude of N2 in parietal area was significantly smaller than that in neutral state. In the late period of working memory updating(400～1 000 ms), the energy in band 13～30 Hz decreased significantly, which was less than that in the neutral state. While no significant difference of updating time was detected between the negative state and neutral state. During the early updating(0～400 ms), the improvement of the energy in band 0～8 Hz was less than that in the neutral state. These results suggested that positive emotion impaired the updating of working memory, while no significant effect of emotion on the updating was observed in the negative states. This research provided experimental support for the improvement of the updating performance of working memory.
		2018 Vol. 37 (3): 297-304 [Abstract] ( 493 ) HTML (1 KB) PDF (2926 KB) ( 634 )

	305	Quantifying the Response of QT Variability to Heart Rate Variability Based on Linear Parametric Model and Information Decomposition Method
		Li Chenxi, Pan Yue, Wang Zhigang, Zhang Zhengguo, Peng Yi
		DOI: 10.3969/j.issn.0258-8021.2018.03.007
		The balance of autonomic nerve system (ANS) plays an important role in avoiding the risk of heart-related diseases. To reveal the regulation of ANS, the response of QT variability (QTV) to heart rate variability (HRV) was analyzed by using linear parametric model in frequency domain and nonlinear information decomposition method. The Holter data were provided by THEW, database Normal was selected as normal controls (Normal, n=186) and database ESRD as typical subjects of ANS dysfunction with high risk for cardiac arrhythmias and sudden cardiac death (ESRD, n=41). 5 min RR interval (RRI) and the corresponding QT interval (QTI) at rest were extracted both in daytime and on night. The QTV fraction related to HRV (LR) in the frequency domain and the predictive information (PI) from RRI to QTI based on the information theory were calculated, combined with time-domain indexes, frequency domain indexes and symbolic dynamic analysis (SDA) of RRI, to explore the possible difference of QTV response to HRV in the two groups and its potential mechanism. There were significant diurnal differences both for LR and PI in Normal, but no significant diurnal variation of that was observed in ESRD, reflecting the loss of ANS reciprocal interaction in ESRD. When comparing the same indexes between Normal group and ESRD group in the same time period, there were no significant differences in LR values in low frequency band between two groups, while LR values in high frequency band in Normal were significant smaller than that in ESRD (Day: 18.36%±17.38% vs 39.37%±23.80%, P<0.05; Night: 28.63%±18.77% vs 42.31%±21.97%, P <0.05); PI on night was significantly higher in Normal compared with that in ESRD (0.310±0.155 vs 0.236±0.131, P <0.05), but no significant difference in PI was found between two groups in daytime. The results demonstrated that linear parametric model and nonlinear prediction based on information decomposition have different sensitivity to ANS activity; The complexity of HRV in regulating QTV in population with high risk for cardiac arrhythmias and sudden cardiac death is reduced.
		2018 Vol. 37 (3): 305-312 [Abstract] ( 305 ) HTML (1 KB) PDF (793 KB) ( 518 )

	313	The Predictive Model for Coronary Artery Lesions in Kawasaki Disease Based on Neural Network
		Zhang Sheng, Tian Jie, Fan Chu, Tan Xuhai, Li Zhe, He Xiangqian
		DOI: 10.3969/j.issn.0258-8021.2018.03.008
		The objective of the study is to find out the risk factors for coronary artery lesions (CAL) in Kawasaki disease (KD) and build the predictive model. The electronic medical record (EMR) data of 1000 KD patients (343 KD with CAL) was collected including the demographic data, laboratory test data, echocardiography and diagnosis data, which were pre-processed for analysis. The risk factors for CAL in KD were selected using association rules. The data set was divided into training set (70%) and testing set (30%), and the neural network (NN) model and logistic regression (LR) model were built. The predictive performance of the two models was evaluated. Results showed that the sensitivity, specificity, accuracy and AUC (Area Under the ROC Curve) of NN model was 0.718, 0.746, 0.737 and 0.796 respectively, which was better than those obtained from LR model[0.175, 0.893, 0.647 and 0.624 respectively]. Thus, the performance of NN model to predict CAL in KD is better than that of LR model.
		2018 Vol. 37 (3): 313-318 [Abstract] ( 472 ) HTML (1 KB) PDF (747 KB) ( 292 )

	319	Research of Force Computation and Physical Simulation of Collision with Friction in Virtual Surgery
		Ou Huitang, Li Jinfang, Mo Jianqing
		DOI: 10.3969/j.issn.0258-8021.2018.03.009
		In the virtual surgery training, the computation of collision force is indispensable. Usually, the friction between colliders is neglected in the computing contact force. This paper proposed a method to compute the contact force with friction. First, the particle system that is widely used in physical simulation in computer graphics was established. And the total ODEs were solved through implicit Euler method. Then, the potential contact points were found out through collision detection and the mathematical models of collision constraints and friction constraints were constructed, and subsequently the contact force with friction was computed by Guass-Seidel-like algorithm. Then 10 single contact and double contact surface contact collision experiments were carried out respectively, and then 100 multiple contact surface collision experiments were carried out, and the average operation time was analyzed. The experimental results showed that the average response time of the model collision was 0.02 s under the contact of single contact surface and double contact surface, and the average fluctuation range was in the range of + 0.02 units after the contact force was stable. In the experiment of the contact of multi contact surface, the average operation time of each contact point was 1.9 ms when 11 contact points contact, which could meet the real time requirement of physical simulation.
		2018 Vol. 37 (3): 319-326 [Abstract] ( 455 ) HTML (1 KB) PDF (3097 KB) ( 318 )

	327	Epileptogenic Zone Identification Based on Nonlinear Interdependence
		Ma Zhen
		DOI: 10.3969/j.issn.0258-8021.2018.03.010
		Exact identification of the epileptogenic zone (EZ) is the basis of epilepsy treatments and helps to reduce side effects. The results of traditional visual methods for identifying the origin of seizures are unsatisfactory in some cases. Signal processing methods could extract substantial information to complement visual inspection in many ways. In this study, EZ identification is regarded as a driver identification problem, and a nonlinear interdependence measure is proposed as an EZ (driver) indicator. It can detect coupling strength and directionality information, especially coupling directionality which can indicate seizure propagation direction, from EEG signals. Two directionally coupled neural mass models are employed for simulation investigation. Two parameters (k and a) can adjust the sensitivity and completeness of proposed interdependence for different applications. Proposed EZ Identification method is also simulated in the context of neural mass models. Simulation results illustrate that proposed EZ identification method can be applied to EZ at different excitatory degree, and achieves an overall identification rate of 98.84% for several EZ types in the cases without synaptic delay and about the same identification rate in the cases with a synaptic delay.
		2018 Vol. 37 (3): 327-334 [Abstract] ( 391 ) HTML (1 KB) PDF (1093 KB) ( 404 )

	335	Research of β-TCP Bone Tissue Engineering Scaffolds Modified with Type I Collagen Based on Three-Dimensional Printing Technique
		Sun Kaiyu, Xu Mingen, Zhou Yongyong
		DOI: 10.3969/j.issn.0258-8021.2018.03.011
		Fabricating individualized tissue engineering scaffolds based on the three-dimensional shape of patient bone defects is required for the successful clinical application of bone tissue engineering.In this work, type I collagen gel was coated on individuated β-TCP scaffolds through 3D printing technique for bone repair.By comparing the influence of filling angle of 0/90°, 0/60°, 0/45°and concentration of coated collagen of 0.10 mg/mL, 0.25 mg/mL, 0.5 mg/mL on the pore diameter, porosity and mechanical properties of β-TCP scaffold, the β-TCP/ collagen scaffold with an optimal filling angle of 0/90° and an optimal concentration of 0.5 mg/mL for coated collagen was chosen, which was able to accurately reproduce the 3D model of design by equipping itself with multilevel pore structure whose mean diameter of megalopore and micropore were 315 μm and 3~5 μm respectively with a porosity of 84%. Meanwhile, due to the compression strength of 12.29±0.88 MPa and elasticity modulus of 116.74±27.75 MPa, it has quite a similarity with adult cancellous bone.In vitro culturing experiments of mouse bone marrow mesenchymal stem cells (mBMSCs)demonstrated that the coated collagen promoted the bioactivity and osteogenic properties, including better cytocompatibility, cell adhesion, proliferation, alkaline phosphatase (ALP) activity, and bone-related gene expressions (Collagen-I, BSP).The results showed that the collagen gel coated β-TCP scaffoldshad the matching shape,good controllable porosity and good osteogenic activity for mBMSCs through 3D printing technique.
		2018 Vol. 37 (3): 335-343 [Abstract] ( 471 ) HTML (1 KB) PDF (9300 KB) ( 216 )

Reviews

	344	Progress of Electromagnetic Detection and Imaging of Magnetic Nanoparticles
		Zhu Jianjian, Yang Wenhui, Wei Shufeng, Wang Zheng, Lv Xing
		DOI: 10.3969/j.issn.0258-8021.2018.03.012
		Over the past decade, magnetic nanoparticles have been developed from the initial MRI contrast agent into a multi-functional material for diagnosis, targeted drug delivery and magneto-hyperthermia. Magnetic nanoparticle imaging (MPI) is a computer tomography technique to construct 3d images by detecting the magnetic properties of superparamagnetic nanoparticles injected into the blood vessels. In addition to angiography and stem cell tracking, magnetic nanoparticle imaging has a range of exciting potential biomedical applications such as real-time fluoroscopy, diagnosis and staging of cancer, in vivo inflammation imaging, temperature display, and functional molecular imaging. MPI attempts to obtain a tracer distribution of measured volumes in a more sensitive, faster and safer way based on existing contrast imaging techniques.In this paper, we first introduced the application of magnetic nanoparticles in molecular imaging and diagnosis, and then introduced the principle of electromagnetic detection of magnetic nanoparticles and the present situation of research as well as problems about system topologies and imaging reconstruction. At the end of this article, we proposed future important trend of the technology.
		2018 Vol. 37 (3): 344-352 [Abstract] ( 523 ) HTML (1 KB) PDF (3932 KB) ( 1372 )

	353	A Review on the Patient Similarity Analysis Based on Electronic Medical Records
		Jia Zheng, Zong Ruijie, Duan Huilong, Li Haomin
		DOI: 10.3969/j.issn.0258-8021.2018.03.013
		Predicting the future health status of a patient has important social and scientific value. At the same time, the accumulating big data in health care domain provides a new basis for obtaining predictive models or establishing predictive methods through medical big data analysis. The patient similarity analysis that provides a general-purpose computer assistant clinical decision support framework based on the predictive knowledge mining from the large practice clinical data generated by a mount of routine patients using the patient distance assessment has paved a way to personalized medicine. Up to date, this method that had been initially approved in many medical domains such as cancer, endocrine diseases and heart diseases, has become a very important direction in clinical translation of artificial intelligence technology. In this paper,the theoretical basis and research progress of the patient similarity analysis were reviewed through introducing the common structure of the patient similarity calculation framework and corresponding key technologies in different processes, such as data preprocessing, dimension reduction, measuring distance of different concepts and generation of similarity. At the same time, existing problems and challenges faced by the patient similarity analysis were proposed.
		2018 Vol. 37 (3): 353-366 [Abstract] ( 624 ) HTML (1 KB) PDF (3136 KB) ( 1179 )

	367	Research Status of Virtual Implantation of Coronary Stent
		Pan Lianqiang, Lin Changyan
		DOI: 10.3969/j.issn.0258-8021.2018.03.014
		Coronary stent implantation is a routine method of interventional therapy for coronary heart disease. How to predict and evaluate the surgical regimen before the surgery in order to optimize and improve has been a clinical concern. With the help of computer platform, the finite element numerical simulation method was used to simulate the virtual implantation of coronary stent, hence the realistic problems were transformed into a mathematical issue, which can be simulated in finite element method. The design of the coronary stent implantation was analyzed and optimized by computer calculation and simulation. Based on the research history and present situation of numerical simulation of coronary stent implantation at home and abroad, this paper introduced the research progress of virtual implantation of coronary stent, and summarizes the research results from the three aspects including the balloon model, vascular wall model and bifurcation model.From the results of these studies, the virtual implantation of coronary stent can assist the clinical selection of the best program, and has significance for leading the development of coronary interventional therapy.
		2018 Vol. 37 (3): 367-371 [Abstract] ( 510 ) HTML (1 KB) PDF (726 KB) ( 385 )

	372	Effects of Fibrous Scaffolds Structure on Cell Behavior
		Li Qiwei, Li Chaojing, Wang Fujun, Ding Wen, Hu Sihan, Wang Lu
		DOI: 10.3969/j.issn.0258-8021.2018.03.015
		Effective strategies to guide cell growth are critical for the development of engineered tissues,and the surface morphology and properties of the biomaterials have appreciable impact on the biocompatibility and biological function, thus the regulation of cell behaviors can provide theoretical supports for developing biocompatible tissue engineering scaffolds. The fibrous scaffolds fabricated by electrospinning could mimic the network structure of the natural extracellular matrix (ECM), supporting cell attachment and guiding tissue formation, therefore fabricating micro- or nanofibers scaffolds are becoming of great interest in the area of biomaterials and tissue engineering. The effect of structural parameters of electrospun scaffolds on the cell behavior was reviewed in this paper, including fiber diameter, pore size, and fiber arrangement. The perspective of fibrous scaffolds and the current improved electrospinning methods for preparing different fibrous structures were also summarized.
		2018 Vol. 37 (3): 372-379 [Abstract] ( 366 ) HTML (1 KB) PDF (1129 KB) ( 635 )

Communications

	380	Research on CSF Shunt Flow Sensor Based on Thermal Conduction
		Li Tongbin, Fang Zhimin, Song Jianhua, Kang Xin
		DOI: 10.3969/j.issn.0258-8021.2018.03.016

		2018 Vol. 37 (3): 380- [Abstract] ( 411 ) HTML (1 KB) PDF (5012 KB) ( 444 )

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