基于深度相机的老年跌倒监护系统

为缓解社会老龄化压力和解决子女照顾老人的精力不足等问题,设计一种基于深度相机的高准确性和低误报性的人体跌倒检测系统。该系统使用RGB相机和红外IR相机获取标定后的老人所在环境的3D图像,并利用深度卷积神经网络提取人体的多个关节点位置,最后基于多个连续帧之间人体关节点的运动变化特征和3D场景特征相结合的方法综合判定老人是否发生跌倒行为。测试实验结果表明该系统能有效地检测到人体的跌倒行为,具有十分优良的鲁棒性。     

基于隐马尔可夫模型的老年人跌倒行为检测方法研究

随着社会老龄化程度的加剧,老年人的安全健康监护需求日益增加。跌倒行为在老年人日常生活中比较常见,它会给老年人带来严重的身体及心理伤害。因此,跌倒检测对于保护老年人的健康及安全具有重要意义。针对跌倒的运动过程,分析人体加速度变化特征,提出基于隐马尔可夫模型(HMM)的跌倒检测方法。将人体跌倒的加速度信号提取为加速度观测序列,并以此为训练样本训练隐马尔可夫模型,建立跌倒过程的概率模型进行跌倒检测。在验证实验中,采集10名志愿者共300例样本,采用5折交叉检验方法,对模型的有效性进行验证。验证结果表明,该方法检测跌倒的准确率为98.2%,灵敏度为91.3%,特异性为99.6%,具有良好的检测效果,可实现对跌倒行为的准确检测。     

基于高速摄像的人体上肢运动信息检测

提出了一种基于高速摄像的人体运动参数检测方法。建立了一套人体上肢运动信息检测系统；设计了快速图像处理算法，准确提取图像中标记点；采用关节瞬心原理对关节角度等参数计算结果进一步修正。所检测出的人体上肢运动参数可以为医疗、体育、仿生学等研究提供参考，所提出的检测方法还可以扩展到人体其它部位运动信息的检测。     

基于指尖视频图像的自适应脉搏信号提取算法研究

针对手指视频图像R分量饱和失真现象,本文提出了一种基于迭代的阈值分割算法,自适应生成R分量待检测区域,通过计算待检测区域灰度均值,从而提取出人体脉搏信号。原始脉搏信号存在基线漂移及高频噪声,结合脉搏信号特征,设计了零相位数字滤波器来滤除噪声干扰。在不同智能手机上采集了指尖视频图像,利用本文提出的算法提取出了待检测区域。考虑到每次测量时指尖压力会有所不同,本文对不同压力下提取的脉搏信号做了对比分析。为了验证本文提出的算法在心率检测方面的准确性,做了心率检测对比实验。结果表明,本文提出的算法能准确提取出人体心率信息,同时具备一定的可移植性,为进一步在智能手机平台上开发生理监测应用提供了一定的理论帮助。     

基于医学图像的人体组织有限元模型的建立方法

在医学图像处理、分析和一些成像算法重构中,建立准确的人体组织轮廓外形和有限元模型对于图像分析和改善重构算法的性能、提高分析和重构结果的准确性具有重要作用。本文使用图像处理与曲线拟合的办法,提取CT图像中的真实人体边界,进行分层的有限元剖分与相关参数设置,获得基于真实人体的有限元重构模型。该方法构建的有限元重构模型已在医学图像处理实际工作中得到了初步的应用。     

基于多传感器数据融合的跌倒检测算法

目的跌倒在老年人生活中是一种常见的现象,是致使老年人发病和死亡的主要原因之一。实时的跌倒检测系统能够及时报警,缩短等待救治的时间,减少由跌倒引起的意外伤害。可是,在大多数的跌倒检测系统中,人们仅利用加速度计设计检测系统,基于单一数据的算法不能完整表征跌倒时身体姿态变化的信息。为此本文拟采用陀螺仪和加速度计的数据设计跌倒检测的算法。方法首先介绍了利用MEMS惯性传感器设计置于腰间的可穿戴的跌倒检测系统,然后对跌倒的规律进行了分析,基于此提出了基于多传感器数据融合的跌倒检测算法,即通过数据融合的技术提取出身体加速度及其动态量和静态量、加速度变化量、身体姿态角、角速度绝对值之和等特征参数,利用多参数设计了基于阈值判定的跌倒检测算法。结果收集10名志愿者做模拟跌倒以及日常活动的数据,对算法的有效性进行验证,取得96.67%的灵敏度和97%的特异性,并且此指标高于Kagans等算法的结果。结论本文提出的算法在跌倒检测中具有较好的有效性及优点。     

多尺度MRI断层图像特征提取新技术研究

由于医学图像本身信号噪声大，边缘呈弱信号特征，用传统的图像边缘检测算法提取图像特征常会将图像中的噪声作为边缘提取，不能准确地反映医学图像中有价值的信息(如病灶大小等)。现提出一种改进算法，此法以现有的小波模极大值特征提取算法为基础，利用模糊理论确定隶属函数，提取弱信号边缘并用多尺度融合理论边缘点合成。结果表明，此方法在提取MRI图像特征的同时，可有效地抑制噪声，有助于剔除图像的伪边缘，准确定位图像边缘信息，有利于图像分割重建，便于医生根据图像确定病灶或组织的位置大小等。     

消化道内窥镜图像异常的人工智能诊断方法研究进展

随着深度学习的出现,图像处理不再局限于人工提取特征,转而对图像进行端到端的预测,实现了人工智能在图像处理领域的又一历史性飞越。作为人工智能医疗领域的热点应用,内镜图像异常检测能够准确快速地筛选整个消化道的异常,为医生提供诊断帮助。该文围绕消化道图像最为常见的息肉、出血、溃疡等异常,对其智能诊断方法展开研究,并探讨机器学习在消化内镜异常检测的应用现状,最后展望了未来消化道内窥镜病灶智能诊断的研究方向。     

基于极限学习机的跌倒检测分类识别研究

由于年龄和身体条件的限制,在老年人群中跌倒是非常普遍的现象。因此,根据老年人跌倒的运动特征,远程监测他们在各个时间段的状态,以便在其摔倒或突发状况时及时采取措施显得尤为重要。针对人体运动状态进行监测,分析人体运动学特征,提出基于极限学习机的跌倒检测算法。运用三维加速度传感器采集人体的三维加速度值,建立跌倒检测特征模型。在此基础上,建立基于极限学习机的跌倒检测分类器,完成对老年人的计算机辅助跌倒检测。实验数据共540例样本,选用了不同数量的训练集和测试集,其中440例作为训练数据,其余100例为测试数据。测试结果表明,准确率为93%,敏感度为87.5%,特异性为91.7%,具有良好的分类性能。在对分类训练的运行时间方面,基于极限学习机的跌倒检测方法与传统的机器学习方法相比具有明显优势。     

建立反映心脏内部各腔室结构的数字化三维模型

背景:心脏内部各腔室精细结构的数字化三维模型,不仅能够加强对心脏生理的认识,还能为心脏电生理仿真与心内膜电生理标测导航的研究提供重要的基础医学数据。目的:建立符合心腔实际解剖结构的三维数据,建立反映心脏内部各腔室结构的数字化三维模型。方法:在MATLAB软件环境下完成图像分割,首先对医学人体断层数据集中心脏切片图像进行自动配准,再运用聚类法根据图像的颜色特征对图像中各组织与成分进行分类,随后经区域生长法处理实现心腔及相关连通区域的图像分割与提取。所得图像导入专用医学图像处理软件,完成心腔三维模型重建。结果和结论:所提出的方法能够重建出较为精细的心腔结构三维模型。此模型中左右心房、心室结构明晰,能清楚地观察到主动脉、上腔静脉等血管,三尖瓣与二尖瓣等微细结构也有所体现。结果说明,重建的三维模型能够准确地反映出人体心腔解剖学结构特征,为心脏电生理方面的仿真和标测研究提供基础的医学数据支持。     

Comparison of protocols for surveillance of methicillin-resistant Staphylococcus aureus (MRSA): medical staff vs ICU patients

To compare the sensitivity of various protocols for methicillin-resistant Staphylococcus aureus (MRSA) surveillance, active surveillance for detecting MRSA nasal colonization was performed on 97 members of the medical staff and 218 patients in the Intensive Care Unit (ICU) of a university hospital. Duplicate nasal swabs were collected from each participant. One was plated directly on a blood agar plate (D-BAP) and observed at 24 and 48 hr. Another was incubated overnight in tryptic soy broth (TSB) with 6.5% NaCl, and subcultured on both BAP (B-BAP) and mannitol salt agar with 4 mg/L of oxacillin (B-MSAOXA). The MRSA colonization rate was similar in the medical staff and patient samples (16.5% vs 11.9%, p = 0.285). Among the medical staff members, the sensitivity of MRSA detection was the same (93.8%) in D-BAP and B-BAP. In the ICU patients, which are a high-risk group, the sensitivity of MRSA detection was improved by adding a pre-enrichment step (73.1% on D-BAP vs 96.2% on B-BAP). The simple direct plating protocol was sufficiently sensitive for the medical staff members, but pre-enrichment was an essential step to increase detection of MRSA in the ICU patients.     

A smart room for hospitalised elderly people: essay of modelling and first steps of an experiment.

We present a study of modelling and the first steps of an experiment of a smart room for hospitalised elderly people. The system aims at detecting falls and sicknesses, and implements four main functions: perception of patient and environment through sensors, reasoning from perceived events and patient clinical findings, action by way of alarm triggering and message passing to medical staff, and adaptation to various patient profiles, sensor layouts, house fixtures and architecture. It includes a physical multisensory device located in the patient's room, and a multi-agent system for fall detection and alarm triggering. This system encompasses a perception agent, and a reasoning agent. The latter has two complementary capacities implemented by sub-agents: deduction of type of alarm from incoming events, and knowledge induction from recorded events. The system has been tested with a few patients in real clinical situation, and the first experiment provides encouraging results which are described in a precise manner.     

Tracking the social dimensions of RFID systems in hospitals

BACKGROUND: Radio frequency identification (RFID) is an emerging technology that is rapidly becoming the standard for hospitals to track inventory, identify patients, and manage personnel. METHODS: Research involved qualitative methods including participant observation and interviews with hospital staff members and industry consultants in the United States. RESULTS: Hospital staff, especially nurses, expressed concern about the surveillance potential of these tracking technologies. Additionally, nursing staff frequently experience an intensification of labor as a result of the implementation of RFID systems because the task of keeping the systems operational often falls upon them. CONCLUSIONS: The social and organizational factors that contribute to the success or failure of RFID systems in hospitals must be further analyzed. The implications of RFID systems, such as privacy concerns and work intensification for nursing and other hospital staff, should be taken into account from the outset, especially during the design and implementation of the technology.     

Evaluation of a fall detector based on accelerometers: A pilot study

As falls and fall-related injuries remain a major challenge in the public health domain, reliable and immediate detection of falls is important so that adequate medical support can be delivered. Available home alarm systems are placed on the hip, but have several shortcomings. A fall detector based on accelerometers and placed at head level was developed, as well as an algorithm able to distinguish between activities of daily living and simulated falls. Accelerometers were integrated into a hearing-aid housing, investigation into the acceleration patterns of the head of a young volunteer during intentional falls. The specificity was assessed by investigation into activities of daily living of the same volunteer. In addition, a healthy elderly woman (83 years) wore the sensor during the day. Three trigger thresholds were identified so that a fall could be recognised: the sum-vector of acceleration in the xy-plane higher than 2 g; the sumvector of velocity of all spatial components right before the impact higher than 0.7ms⁻¹; and the sum-vector of acceleration of all spatial components higher than 6g. The algorithm was able to discriminate activities of daily living from intentional falls. Thus high sensitivity and specificity of the algorithm could be demonstrated that was better than in other fall detectors worn at the hip or wrist at the same stage of development.     

Medical safety management in the blood collection center of clinical laboratory

Safety management is essential for providing patients with medical services. Our hospital, opened at Kashiwa in 1987, has been building up systems and taking a number of steps to reduce the blood collection related problems, including venipuncture-related infection, nerve injury, and vasovagal reflex with syncope in accordance with guidelines for the standard method of venipuncture blood collection. We also have made efforts for improving medical services, including reductions in patient waiting time and prevention of patient misidentification, medical test malpractice, and patient privacy. However, ultimately, it is of obvious significance to educate and train communication skills for humanity and friendly kindness because most of medical accidents are basically attributed to communication errors between patients and medical staff.     

Identitovigilance et intégration de l’informatique transfusionnelle dans les établissements de santé

In hospitals, patients’ identity mismatch is the cause of serious incidents during transfusion or other healthcare processes. With a wider use of healthcare software, new dysfunctions and risks may occur in case of ID mismatch between different software and for transfusion between blood bank and hospital. Four critical stages are to be considered when using software management of hospital care: initial patients’ identification, modifications of identification data during hospital stay and then synchronization of all different software, proper use of patient’s identification means by hospital staff and share of patient’s identification data with others such as the Établissement français du sang (EFS) for transfusional purposes. Perpignan Hospital has put in place an identity vigilance group whose role is to enforce a safe patients’ identification policy. It has to inform and train hospital staff, make sure that all software requiring patients’ ID are correctly synchronized, especially when sharing data with EFS and monitor improvement with a surveillance system. As transfusion care is one of the more patients’ ID sensitive, particularly when computer assisted, haemovigilance officers are highly involved in identity vigilance.     

Electrical safety in the hospital—1974

The topic of electrical safety in the hospital has been of considerable interest since it covered that certain catheterized patients can be accidentally electrocuted by currents well below the perceivable level. Prompted by a safety scare that began in 1969, hospitals and manufacturers of medical equipment have since become safety conscious and have taken steps to effectively reduce the safety hazard to both patients and hospital staff. This paper is intended to provide the reader with a basic understanding of electrical safety problems in the hospital. The phisiological effects of electricity in the body, electrical safety hazards in the hospital and methods of protecting patients from both macroshock and microshock hazards are described and discussed. The paper concludes with recommendations for a relatively inexpensive, but effective and up-to-date electrical safety program for the hospital.     

Système de télésurveillance sonore pour la détection de situations de détresse

The telemedicine and medical remote monitoring in particular, today represents an effective solution to the health professional shortcomings facing to the increasing older population. In addition to the comfort of being at home, this system decreases the cost of long hospitalization. The proposed system achieves the home medical telesurveillance by means of microphones and medical sensors to detect a distress situation. This system is designed for elderly people at home or for the patient with chronic illness. The experimental apartment is equipped with medical sensors (tensiometer, oxymeter, balance, etc.), infrared position sensors, and acoustic sensors (1 microphone/room). The originality of the system comes from the replacement of the video surveillance with a sound surveillance. The sound analysis system is divided in two stages: firstly, the detection/extraction of the sounds is operated and secondly, a classification of these sounds in known classes takes place. A wavelet-based algorithm with good performance when applied in noisy environments is proposed. The acoustical classification step uses a Gaussian Mixture Models to classify the sounds according to the 7 predefined classes. The detection algorithm allows an equal error rate of 0% for the signal to noise ratio superior or equal to 10 dB and 4% for the 0 dB. The proposed system coupled with a classical medical telesurveillance system will bring extra information needed for the reliability of the global system.     

Sensor technologies aiming at fall prevention in institutionalized old adults: A synthesis of current knowledge

BackgroundFalls are a serious health problem in old adults especially in nursing home residents and hospitalized patients. To prevent elderly from falling, sensors have been increasingly used in intramural care settings. However, there is no clear overview of the current used technologies and their results in fall prevention.ObjectivesThe present study reviews sensor systems that prevent falls in geriatric patients living in an intramural setting and describe fall rates, fall-related injuries, false alarms, and user experience associated with such systems.MethodsWe conducted a systematic search for studies that used sensor technologies with the aim to prevent falls in institutionalized geriatric patients.ResultsA total of 12 studies met the search criteria. Three randomized clinical trials reported no reductions in fall rate but three before–after studies reported significant reductions of 2.4–37 falls per 1000 patient days. Although there was up to 77% reduction in fall-related injuries and there was relatively low, 16%, rate of false alarms, the current data are inconsistent whether current sensor technologies are effective in reducing the number of falls in institutionalized geriatric patients. The occurrence of false alarms (16%) was too high to maintain full attention of the nursing staff. Additionally including the users opinion and demands in developing and introducing sensor systems into intramural care settings seems to be required to make an intervention successful.ConclusionThe evidence is inconsistent whether the current sensor systems can prevent falls and fall-related injuries in institutionalized elderly. Further research should focus more comprehensively on user requirements and effective ways using intelligent alarms.