Fitting a bilateral transhumeral amputee with utensil prostheses and their functional assessment 10 years later: a case report

Traumatic transhumeral amputations resulting from electric burn injury are uncommon and present a significant rehabilitation challenge. Compensating for loss of fine, coordinated function of the upper extremities with prostheses is difficult medically, technologically, psychologically, and socially. We followed up a patient with traumatic bilateral transhumeral amputation who was fitted with specially designed bilateral low-temperature utensil prostheses for 10 years. A bilateral utensil prosthesis consists of 2 thermoplastic sockets, an elastic harness, 2 utensil holders, and several different utensils. The characteristics of utensil prostheses are low cost, quick fabrication, and responsiveness to a patient's needs. Ten years after the patient's first prosthesis fitting, he still used these specially designed prostheses. By using these devices and his feet, the patient has regained independence in most activities of daily living and gained a new working skill.     

Data-driven prediction and control of wastewater treatment process through the combination of convolutional neural network and recurrent neural network

It is widely believed that effective prediction of wastewater treatment results (WTR) is conducive to precise control of aeration amount in the wastewater treatment process (WTP). Conventional biochemical mechanism-driven approaches are highly dependent on complicated and redundant model parameters, resulting in low efficiency. Besides, sharp increase in business volume of wastewater treatment requires automatic operation technologies for this purpose. Under this background, researchers started to introduce the idea of data mining to model the WTP, in order to automatically predict WTR given inlet conditions and aeration amount. However, existing data-driven approaches for this purpose focus on modelling of the WTP at independent timestamps, neglecting sequential characteristics of timestamps during the long-term treatment process. To tackle the challenge, in this paper, a novel prediction and control framework through combination of convolutional neural network (CNN) and recurrent neural network (RNN) is proposed for prediction of the WTR. Firstly, the CNN model is utilized to automatically extract the local features of each independent timestamp in the WTP and make them encoded. Next, the RNN model is employed to represent global sequential features of the WTP on the basis of local feature encoding. Finally, we conduct a large number of experiments to verify efficiency and stability of the proposed prediction framework.

This work proposes a novel data-driven mechanism for prediction of wastewater treatment results through mixture of two neural network models.     

Self-organized neural network for the quality control of 12-lead ECG signals

Telemedicine is very important for the timely delivery of health care to cardiovascular patients, especially those who live in the rural areas of developing countries. However, there are a number of uncertainty factors inherent to the mobile-phone-based recording of electrocardiogram (ECG) signals such as personnel with minimal training and other extraneous noises. PhysioNet organized a challenge in 2011 to develop efficient algorithms that can assess the ECG signal quality in telemedicine settings. This paper presents our efforts in this challenge to integrate multiscale recurrence analysis with a self-organizing map for controlling the ECG signal quality. As opposed to directly evaluating the 12-lead ECG, we utilize an information-preserving transform, i.e. Dower transform, to derive the 3-lead vectorcardiogram (VCG) from the 12-lead ECG in the first place. Secondly, we delineate the nonlinear and nonstationary characteristics underlying the 3-lead VCG signals into multiple time-frequency scales. Furthermore, a self-organizing map is trained, in both supervised and unsupervised ways, to identify the correlations between signal quality and multiscale recurrence features. The efficacy and robustness of this approach are validated using real-world ECG recordings available from PhysioNet. The?average performance was demonstrated to be 95.25% for the training dataset and 90.0% for the independent test dataset with unknown labels.     

At the interface: convergence of neural regeneration and neural prostheses for restoration of function.

The rapid pace of recent advances in development and application of electrical stimulation of the nervous system and in neural regeneration has created opportunities to combine these two approaches to restoration of function. This paper relates the discussion on this topic from a workshop at the International Functional Electrical Stimulation Society. The goals of this workshop were to discuss the current state of interaction between the fields of neural regeneration and neural prostheses and to identify potential areas of future research that would have the greatest impact on achieving the common goal of restoring function after neurological damage. Identified areas include enhancement of axonal regeneration with applied electric fields, development of hybrid neural interfaces combining synthetic silicon and biologically derived elements, and investigation of the role of patterned neural activity in regulating various neuronal processes and neurorehabilitation. Increased communication and cooperation between the two communities and recognition by each field that the other has something to contribute to their efforts are needed to take advantage of these opportunities. In addition, creative grants combining the two approaches and more flexible funding mechanisms to support the convergence of their perspectives are necessary to achieve common objectives.     

Augmented auditory information in the control of upper-limb prostheses

The present study compared the effects of augmented auditory information on the linear positioning performance of individuals with natural or prosthetic limbs. Subjects were ten male volunteers, five of whom had above-elbow amputations and had used their prosthetic devices for an average of 8.3 years. The other five subjects were selected from a volunteer pool. Movements were made and measured on a standard linear slide whose cursor had a pulsing infrared diode attached opposite the subject. An infrared camera and microprocessor translated diode movement into a corresponding change in voltage. The voltage was simultaneously applied to an audio device which supplied the augmented feedback. The movement of the cursor by the subject was paralleled by a linearly-related change in audible frequency (Hz). The subjects performed 15 trials at each of three retentions (immediate, 15-sec filled, and 15 sec unfilled), both with and without the augmented feedback, for a total of 90 trials. The results of group X feedback X retention intervals analysis of variance on absolute and variable error indicated both a group X retention and a group X feedback interaction. Subjects using a prosthetic limb to produce the movement were less accurate and more variable than the "normal" subjects when augmented feedback was not concurrent with response.     

Performance of two neural network models in bathymetry

Neural networks are widely used as predictors in several fields of applications, such as prediction of shallow water depth. The purpose of this study is to investigate the performance of two artificial neural networks models as potential methods in bathymetry. A comparison approach is used to evaluate network models, the regression tree and an inversion model. The high-resolution IKONOS and moderate-resolution Landsat satellite images serve as the case studies, and results based on the root mean square errors and coefficient of determination (R²) show that artificial neural networks outperform the inversion model and the regression tree.     

Use of two-axis joystick for control of externally powered shoulder disarticulation prostheses

We explored a new method for simple and accurate control of shoulder movement for externally powered shoulder disarticulation prostheses with a two-axis joystick. We tested 10 subjects with intact shoulders and arms to determine the average amount of shoulder motion and force available to control an electronic input device. We then applied this information to two different input strategies to examine their effectiveness: (1) a traditional rocker potentiometer and a pair of force-sensing resistors and (2) a two-axis joystick. Three nondisabled subjects and two subjects with shoulder disarticulation amputations attempted to control an experimental externally powered shoulder using both control strategies. Two powered arms were tested, one with powered flexion/extension and humeral rotation and one with powered flexion/extension and adduction/abduction. Overwhelmingly, the subjects preferred the joystick control, because it was more intuitively linked with their shoulder movement. Additionally, two motions (one in each axis) could be controlled simultaneously. This pilot study provides valuable insight into an effective means of controlling high-level, externally powered prostheses with a two-axis joystick.     

Initial stability of a percutaneous osseointegrated endoprosthesis with proximal interlocking screws for transhumeral amputees

BackgroundPercutaneous osseointegrated devices for skeletal fixation of prosthetic limbs have the potential to improve clinical outcomes in the transhumeral amputee population. Initial endoprosthesis stability is paramount for long-term osseointegration and safe clinical introduction of this technology. We evaluated an endoprosthetic design featuring a distally porous coated titanium stem with proximal slots for placement of bicortical interlocking screws.MethodsYield load, ultimate failure load, and construct stiffness were measured in 18 pairs of fresh-frozen and thawed cadaver humeri, at distal and proximal amputation levels, without and with screws, under axial pull-out, torsion, and bending loads. Paired statistical comparisons were performed without screws at the two resection levels, and at distal and proximal levels with and without screws.FindingsWithout screws, the location of the amputation influenced the stability only in torsional yield (p = 0.032) and torsional ultimate failure (p = 0.033). Proximally, the torsional yield and the torsional ultimate failure were 44% and 47% of that distally. Screws improved stability. In axial pull-out, screws increased the distal ultimate failure 3.2 times (p = 0.003). In torsion, screws increased the yield at the proximal level 1.9 times (p = 0.035), distal ultimate failure load 3.3 times (p = 0.016) and proximal ultimate failure 6.4 times (p = 0.013). In bending, screws increased ultimate failure at the proximal level 1.6 times (p = 0.026).InterpretationProximal slots and bicortical interlocking screws may find application in percutaneous osseointegrated devices for patients with amputations, especially in the less stable proximal bone of a short residual limb.     

Temperature measurement and control system for transtibial prostheses: Functional evaluation

The accumulation of heat inside the prosthetic socket increases skin temperature and fosters perspiration, which consequently leads to high tissue stress, friction blister, discomfort, unpleasant odor, and decreased prosthesis suspension and use. In the present study, the prototype of a temperature measurement and control (TM&C) system was designed, fabricated, and functionally evaluated in a phantom model of the transtibial prosthetic socket. The TM&C system was comprised of 12 thermistors divided equally into two groups that arranged internal and external to a prosthetic silicone liner. Its control system was programmed to select the required heating or cooling function of a thermal pump to provide thermal equilibrium based on the amount of temperature difference from a defined set temperature, or the amount of difference between the mean temperature recorded by inside and outside thermistors. A thin layer of aluminum was used for thermal conduction between the thermal pump and different sites around the silicone liner. The results showed functionality of the TM&C system for thermoregulation inside the prosthetic socket. However, enhancing the structure of this TM&C system, increasing its thermal power, and decreasing its weight and cost are main priorities before further development.     

Boundary-aware context neural network for medical image segmentation

Medical image segmentation can provide a reliable basis for further clinical analysis and disease diagnosis. With the development of convolutional neural networks (CNNs), medical image segmentation performance has advanced significantly. However, most existing CNN-based methods often produce unsatisfactory segmentation masks without accurate object boundaries. This problem is caused by the limited context information and inadequate discriminative feature maps after consecutive pooling and convolution operations. Additionally, medical images are characterized by high intra-class variation, inter-class indistinction and noise, extracting powerful context and aggregating discriminative features for fine-grained segmentation remain challenging. In this study, we formulate a boundary-aware context neural network (BA-Net) for 2D medical image segmentation to capture richer context and preserve fine spatial information, which incorporates encoder-decoder architecture. In each stage of the encoder sub-network, a proposed pyramid edge extraction module first obtains multi-granularity edge information. Then a newly designed mini multi-task learning module for jointly learning segments the object masks and detects lesion boundaries, in which a new interactive attention layer is introduced to bridge the two tasks. In this way, information complementarity between different tasks is achieved, which effectively leverages the boundary information to offer strong cues for better segmentation prediction. Finally, a cross feature fusion module acts to selectively aggregate multi-level features from the entire encoder sub-network. By cascading these three modules, richer context and fine-grain features of each stage are encoded and then delivered to the decoder. The results of extensive experiments on five datasets show that the proposed BA-Net outperforms state-of-the-art techniques.     

A shared neural network for simple reaction time

Simple reaction time, a simple model of sensory-to-motor behavior, has been extensively investigated and its role in inferring elementary mental organization has been postulated. However, little is known about the neuronal mechanisms underlying it. To elucidate the neuronal substrates, functional magnetic resonance imaging (fMRI) signals were collected during a simple reaction task paradigm using simple cues consisting of different modalities and simple triggered movements executed by different effectors. We hypothesized that a specific neural network that characterizes simple reaction time would be activated irrespective of the input modalities and output effectors. Such a neural network was found in the right posterior superior temporal cortex, right premotor cortex, left ventral premotor cortex, cerebellar vermis, and medial frontal gyrus. The right posterior superior temporal cortex and right premotor cortex were also activated by different modality sensory cues in the absence of movements. The shared neural network may play a role in sensory triggered movements.     

Application of artificial neural network to fMRI regression analysis

We used an artificial neural network (ANN) to detect correlations between event sequences and fMRI (functional magnetic resonance imaging) signals. The layered feed-forward neural network, given a series of events as inputs and the fMRI signal as a supervised signal, performed a non-linear regression analysis. This type of ANN is capable of approximating any continuous function, and thus this analysis method can detect any fMRI signals that correlated with corresponding events. Because of the flexible nature of ANNs, fitting to autocorrelation noise is a problem in fMRI analyses. We avoided this problem by using cross-validation and an early stopping procedure. The results showed that the ANN could detect various responses with different time courses. The simulation analysis also indicated an additional advantage of ANN over non-parametric methods in detecting parametrically modulated responses, i.e., it can detect various types of parametric modulations without a priori assumptions. The ANN regression analysis is therefore beneficial for exploratory fMRI analyses in detecting continuous changes in responses modulated by changes in input values.     

采用人工神经网络对肌肉痉挛进行定量评估

利用系统辨识方法研究脊髓多节段硬化病人与正常对照组的髌韧带反射特性 ,抽取一系列系统特征参数。在此基础上 ,探讨利用人工神经网络对测量的系统参数进行分级和评定的方法及效果 ,证明这种方法对腱反射特征的定量评估是行之有效的     

BP神经网络在疾病分析影响因素中的作用

背景:疾病产生原因复杂多样,临床医生对大量样本病历数据挖掘的探讨往往缺乏有效的手段,信息技术的应用能力有待提高.目的:利用人工神经网络的BP算法,对临床大样本量的病历进行分析,以找出某种疾病的致病因素与疾病本身之间的内在关系.方法:以高血压病为例,以某中医院2010-07的高血压患者病历数据为实验数据,对疾病的影响因素进行建模,优选 Microsoft SQL Server 2005 Analysis Services智能工具,分析其挖掘结果,并利用单独查询进行预测与决策支持.结果与结论:应用基于BP算法的人工神经网络分析疾病的致病因素对疾病本身的影响有较好的预测效果,有利于提升医务人员借助信息技术方法在临床诊断的水平,提高疾病诊断效率.

Abstract：

BACKGROUND: Disease pathogenic factors are complicated. There is not an effective method to analyze large sample data mining, and application ability of information technology of clinical doctors needs to be improved. OBJECTIVE: Using BP algorithm of artificial neural network to analyze large sample clinical cases, in order to explore inner relations between disease pathogenic factors and diseases.METHODS: Take hypertension for example, medical data of patients with hypertension in a traditional Chinese medical hospital served as experimental data, and the influence factors of the disease were simulated with Microsoft SQL Server 2005 Analysis Services, the mining data was analyzed, and a single query was used as prediction and decision support.RESULTS AND CONCLUSION: Analysis of effect of disease pathogenic factors on disease itself based on artificial neural network with BP algorithm has good predictive effect in clinical diagnosis, which is of benefit to enhance the diagnostic efficiency of medical personnel using information technology.     

卷积神经网络在骨骼肌肉放射学中的研究进展

人工智能逐渐应用于医学图像诊断，显现出良好的效率和诊断准确率。卷积神经网络（CNN）是人工智能领域的一项突破，在某些方面已可与放射科医师相媲美，展现出巨大的临床应用前景。本文就CNN在骨骼肌肉放射学中的研究进展进行综述。     

人工神经网络在肌肉痉挛分类中的应用

肌肉痉挛 (ｓｐａｓｔｉｃｉｔｙ)是一种伴随着上运动神经损伤 ,如脊髓损伤、脊髓多节段硬化、中风以及帕金森氏病等出现的一种并发症 ,其表现主要包括肌张力升高、肌肉僵硬、抽搐性痉挛等。目前对肌肉痉挛的评估尚无好方法 ,评价标准很不统一 ,影响了对疾病的准确诊断和疗效判定。在临床和生物医学工程领域 ,人们一直在探讨对肌肉痉挛程度的客观评价指标和方法 ,目前常用的评估方法包括 :临床分级测评、摆动实验、肌肉反射与腱反射等 ,其中临床分级测评应用得较广泛。本研究利用系统辨识方法研究脊髓多节段硬化患者与正常对照组的髌韧带反…