WO2019100417A1 - Ecg信号的并行分析装置、方法和移动终端 - Google Patents

Ecg信号的并行分析装置、方法和移动终端 Download PDF

Info

Publication number
WO2019100417A1
WO2019100417A1 PCT/CN2017/113408 CN2017113408W WO2019100417A1 WO 2019100417 A1 WO2019100417 A1 WO 2019100417A1 CN 2017113408 W CN2017113408 W CN 2017113408W WO 2019100417 A1 WO2019100417 A1 WO 2019100417A1
Authority
WO
WIPO (PCT)
Prior art keywords
memory
data
ecg signal
cpu
gpu
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2017/113408
Other languages
English (en)
French (fr)
Inventor
李烨
樊小毛
蔡云鹏
姚启航
杨玉洁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to US15/739,719 priority Critical patent/US10743789B2/en
Priority to EP17933117.8A priority patent/EP3716134B1/en
Publication of WO2019100417A1 publication Critical patent/WO2019100417A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/332—Portable devices specially adapted therefor
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/30—Input circuits therefor
    • A61B5/307—Input circuits therefor specially adapted for particular uses
    • A61B5/308—Input circuits therefor specially adapted for particular uses for electrocardiography [ECG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A61B5/35—Detecting specific parameters of the electrocardiograph cycle by template matching
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A61B5/366—Detecting abnormal QRS complex, e.g. widening
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
    • A61B5/0004—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by the type of physiological signal transmitted
    • A61B5/0006—ECG or EEG signals
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • A61B5/024—Measuring pulse rate or heart rate
    • A61B5/0245—Measuring pulse rate or heart rate by using sensing means generating electric signals, i.e. ECG signals
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/333—Recording apparatus specially adapted therefor
    • A61B5/335—Recording apparatus specially adapted therefor using integrated circuit memory devices
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising

Definitions

  • the present disclosure relates to the field of ECG signal processing technologies, and in particular, to a parallel analysis apparatus, method, and mobile terminal for ECG signals.
  • ECG Electrocardiogram
  • Heart rate, rhythm disorder, or morphological changes of ECG signals may become pathological indications
  • the ECG waveform can detect a variety of heart diseases such as myocardial infarction, cardiomyopathy, and myocarditis.
  • an object of the present disclosure is to provide a parallel analysis apparatus, method, and mobile terminal for ECG signals to improve the analysis efficiency of an ECG signal, thereby improving the timeliness of analysis feedback of an ECG signal.
  • the present disclosure provides a parallel analysis device for an ECG signal, the device comprising: an integrated memory, a CPU and a GPU, the integrated memory including a first memory for the CPU and a second memory for the GPU, and the CPU is accessible The second memory; the CPU and the GPU transmit data through the integrated memory; the CPU is configured to perform first-level noise reduction processing on the received ECG original signal to obtain a first-level ECG signal; and perform abnormal heartbeat on the feature data extracted by the GPU.
  • the classification is initially screened to obtain suspected abnormal heart beat data; the GPU is used for feature extraction of the primary ECG signal to obtain feature data; and for performing secondary noise reduction processing on the primary ECG signal to obtain secondary ECG signals, application
  • the template matching classification method processes the suspected abnormal heart beat data and the secondary ECG signal to obtain the final abnormal heart beat data.
  • an embodiment of the present disclosure provides a first possible implementation manner of the first aspect, wherein the CPU includes an original signal receiving module, configured to receive an ECG original signal, and store the ECG original signal in the first memory; a first pre-processing module, configured to load the ECG original signal of the first memory to perform a first-level noise reduction process, obtain a first-level ECG signal, and store the first-level ECG signal in the second memory; the first abnormal heart beat classification module is configured to: The feature data is acquired according to the storage location information, and the characteristic data is subjected to abnormal heartbeat classification according to the set rule determination manner, and the suspected abnormal heartbeat data is obtained, and the suspected abnormal heartbeat data is stored in the second memory.
  • the disclosure provides a second possible implementation manner of the first aspect, wherein the GPU includes: a feature detection module, configured to load a level from the second memory The ECG signal is subjected to feature extraction to obtain feature data; the feature data is stored in the second memory; the second pre-processing module is configured to load the first-level ECG signal for secondary noise reduction processing to obtain the secondary ECG signal; and the second abnormal heart beat a classification module, configured to acquire a secondary ECG signal from the second preprocessing module, and obtain a suspected abnormal heartbeat data from the second memory, according to the set The template matching method and the secondary ECG signal reconfirm the suspected abnormal heart beat data to obtain the final abnormal heart beat data, and store the final abnormal heart beat data in the second memory.
  • a feature detection module configured to load a level from the second memory
  • the ECG signal is subjected to feature extraction to obtain feature data
  • the feature data is stored in the second memory
  • the second pre-processing module is configured to load the first-level ECG signal for secondary noise reduction processing to obtain the secondary E
  • the disclosure provides a third possible implementation manner of the first aspect, wherein the integrated memory includes: a mapping module, configured to use feature data and a final abnormality The storage location information of the beat data is mapped to the first memory, so that the CPU acquires the corresponding data according to the storage location information.
  • a mapping module configured to use feature data and a final abnormality The storage location information of the beat data is mapped to the first memory, so that the CPU acquires the corresponding data according to the storage location information.
  • an embodiment of the present disclosure provides a fourth possible implementation of the first aspect, wherein the first pre-processing module includes an IIR filter for the ECG original signal Performing a filtering process to obtain a first-order ECG signal; the second pre-processing module includes an artifact removal unit for performing artifact removal processing on the primary ECG signal to obtain a secondary ECG signal.
  • the embodiment of the present disclosure provides the fifth possible implementation manner of the first aspect, wherein the feature detection module includes: a morphological transformation unit, configured for the primary ECG The signal is transformed, and the morphological form of the ECG signal is output; the R wave detecting unit is configured to perform R wave detection on the morphological form of the ECG signal, and output the detection result; the QRS group detection is used to perform QRS complex detection on the detection result. Outputs feature data containing QRS complexes.
  • the present disclosure provides a parallel analysis method for an ECG signal, the method being applied to a mobile terminal, the mobile terminal comprising: an integrated memory, a CPU and a GPU, and the integrated memory includes a first memory for the CPU and used by the GPU a second memory, and the CPU can access the second memory; the CPU and the GPU transmit data through the integrated memory; the method includes: the CPU performs first-level noise reduction processing on the received ECG original signal to obtain a first-level ECG signal; ECG signal is extracted to obtain feature data; the CPU performs abnormal heartbeat classification and initial screening processing on the feature data to obtain suspected abnormal heartbeat data; the GPU performs secondary noise reduction on the first-level ECG signal to obtain the secondary ECG signal; The template matching classification method is used to process the suspected abnormal heart beat data and the secondary ECG signal to obtain the final abnormal heart beat data.
  • an embodiment of the present disclosure provides a first possible implementation manner of the second aspect, wherein the method further includes: the CPU acquiring the final abnormal heart rate data, and uploading the final abnormal heart rate data to the remote medical center. Platform; the CPU receives a report from the medical platform based on the feedback of the final abnormal heart rate data.
  • the embodiment of the present disclosure provides the second possible implementation manner of the second aspect, wherein the CPU and the GPU transmit data through the integrated memory, including: integrated memory
  • the data stored in the first memory by the CPU is copied to the second memory, and the storage location information of the data stored in the second memory by the GPU is mapped to the first memory.
  • the present disclosure provides a mobile terminal including the parallel analysis device of the ECG signal described above.
  • the integrated memory includes a first memory for the CPU and a second memory for the GPU; and the first-level noise reduction of the ECG original signal by the CPU Processing, and performing abnormality heartbeat classification and screening processing on the feature data extracted by the GPU; performing feature extraction on the first-order ECG signal obtained by the first-stage noise reduction processing by the GPU, and performing second-level noise reduction processing on the first-level ECG signal.
  • the suspected abnormal heart beat data obtained by the primary screening process and the secondary ECG signal obtained by the secondary noise reduction processing are processed to obtain the final abnormal heart beat data.
  • the CPU and the GPU jointly process various tasks in the ECG signal analysis process, and the GPU performs complex computation tasks in a parallel processing manner, thereby improving the analysis efficiency of the ECG signal, thereby improving the timeliness of the analysis feedback of the ECG signal. At the same time, it reduces the power consumption of the device and improves the user experience.
  • the manner of data copying and mapping between the above-mentioned memories can avoid a large amount of data transmission and reduce the transmission time of data from the CPU end to the GPU end, and the method of data transmission through the bus and the communication line is further reduced.
  • the analysis efficiency of the ECG signal is improved.
  • FIG. 1 is a schematic structural diagram of a parallel analysis apparatus for an ECG signal according to an embodiment of the present disclosure
  • FIG. 2 is a schematic structural diagram of another parallel analysis apparatus for ECG signals according to an embodiment of the present disclosure
  • FIG. 3 is a flowchart of a parallel analysis method of an ECG signal according to an embodiment of the present disclosure
  • FIG. 4 is a flowchart of another method for parallel analysis of ECG signals according to an embodiment of the present disclosure.
  • Long-term ECG can be used to aid in the diagnosis of heart disease such as intermittent arrhythmias.
  • the user can collect the ECG signal through the wearable cardiac monitoring device, and send the ECG signal to the cloud platform connected to the device; the cloud platform analyzes and diagnoses the ECG signal, and then feeds the diagnosis result to the monitoring device, or On the user's mobile terminal. Since the amount of data of the ECG signal is large, and the cloud platform may continuously receive a large number of ECG signals sent by the user, the manner in which the cloud platform processes the ECG signal imposes a large computational burden on the cloud processor, resulting in ECG. The timeliness and reliability of signal processing feedback are not guaranteed.
  • the task of analyzing and diagnosing ECG signals can be completed by mobile terminals such as wearable cardiac monitoring devices, mobile phones, and tablet computers; however, due to limited CPU performance of mobile terminals, it is still difficult to afford long-term ECG signal processing, and timely At the same time, the processing process needs to consume a large amount of power of the device, and the battery loss is large for a mobile terminal with limited battery capacity.
  • the embodiment of the present disclosure provides a parallel analysis device, method, and mobile terminal for ECG signals; the technology can be applied to mobile devices such as wearable cardiac monitoring devices, mobile phones, and tablets. In the terminal, it is used to assist in the diagnosis of heart diseases such as intermittent arrhythmia.
  • the technique can be implemented in related software or hardware, which is described below by way of implementation.
  • a schematic structural diagram of a parallel analysis device for an ECG signal includes: an integrated memory 10 , a CPU 11 , and a GPU (Graphic Processing Unit) 12 .
  • the integrated memory 10 includes a CPU for use by the CPU 11 .
  • the CPU 11 is configured to perform a first-level noise reduction process on the received ECG original signal to obtain a first-level ECG signal, and perform an abnormal heart rate classification preliminary screening process on the feature data extracted by the GPU 12 to obtain suspected abnormal heart beat data;
  • the GPU 12 is configured to perform feature extraction on the first-level ECG signal to obtain feature data, and perform second-level noise reduction processing on the first-level ECG signal to obtain a secondary ECG signal, and apply a template matching classification method to the suspected abnormal heart beat data and two.
  • the ECG signal is processed to obtain the final abnormal heart rate data.
  • the above CPU and GPU are respectively used to perform different tasks in ECG signal analysis, and some tasks can be executed in parallel; for example, when the CPU performs abnormal heartbeat classification initial screening processing, the GPU can perform secondary noise reduction processing on the first-level ECG signal.
  • the task with complex computational complexity is weak, and the GPU is a multi-core processing method, which can process tasks with complex computational complexity such as image calculation in parallel, or have inherent parallel features.
  • Algorithm according to the attributes of the task, reasonable allocation of the processor to be executed can improve the efficiency of ECG signal analysis; for example, the above feature extraction step, which usually requires image recognition, calculation, etc., and the calculation amount is large, and the step is performed by the GPU. Execution can greatly improve the efficiency of ECG signal analysis.
  • the integrated memory may be implemented by a memory chip, wherein the first memory and the second memory may be divided by software; the first memory may be used to store data such as an ECG original signal, a primary ECG signal, and a suspected abnormal heartbeat data; The second memory may be used to store data such as primary ECG signals, feature data, suspected abnormal heart beat data, and final abnormal heart rate data.
  • the first memory and the second memory can be transferred by copying and mapping; for example, the integrated memory can copy the data stored by the CPU in the first memory to the second memory, and the data stored in the second memory by the GPU.
  • the storage location information is mapped to the first memory.
  • the CPU can access the first memory or the second memory, when the CPU needs to acquire the data in the second memory, only the storage address of the data in the second memory needs to be mapped to the first a memory, the CPU accesses the second memory according to the mapped storage address, and then obtains corresponding Data;
  • the GPU can only access the second memory, when the GPU needs to acquire the data in the first memory, the integrated chip needs to copy the data from the first memory to the second memory, and then read by the GPU.
  • the integrated memory includes a first memory for use by the CPU and a second memory for use by the GPU; performing first-level noise reduction processing on the original ECG signal by the CPU, and The feature data extracted by the GPU is subjected to initial screening processing of abnormal heartbeat classification; the first-level ECG signal obtained by the first-level noise reduction processing is extracted by the GPU, and the second-level noise reduction processing is performed on the primary ECG signal, and then the template is applied.
  • the matching classification method processes the suspected abnormal heart beat data obtained by the primary screening process and the secondary ECG signal obtained by the secondary noise reduction processing to obtain the final abnormal heart beat data.
  • the CPU and the GPU jointly process various tasks in the ECG signal analysis process, and the GPU performs complex computation tasks in a parallel processing manner, thereby improving the analysis efficiency of the ECG signal, thereby improving the timeliness of the analysis feedback of the ECG signal. At the same time, it reduces the power consumption of the device and improves the user experience.
  • the manner of data copying and mapping between the above-mentioned memories can avoid a large amount of data transmission and reduce the transmission time of data from the CPU end to the GPU end, and the method of data transmission through the bus and the communication line is further reduced.
  • the analysis efficiency of the ECG signal is improved.
  • FIG. 2 a schematic structural diagram of another parallel analysis device for ECG signals is implemented; the device is implemented on the basis of the device shown in FIG. 1; the device includes: an integrated memory 10, a CPU 11 and a GPU 12, and the integrated memory 10 includes The first memory 101 used by the CPU 11 and the second memory 102 for use by the GPU 12, and the CPU 11 can access the second memory 102; the CPU 11 and the GPU 12 transmit data through the integrated memory 10.
  • Mobile terminals such as smart phones have highly integrated circuits that combine major components (such as CPU, GPU, memory, etc.) into a single chip; this method enables high-bandwidth data transmission; at the same time, ultra-wideband memory metrics accelerate memory and Data transfer speed between CPU/GPU; in addition, The CPU and GPU memory are integrated on the same chip and separated by embedded software. During the execution of the task, the task will be transferred. Therefore, memory mapping technology can be introduced to map the same physical memory to the memory space of the CPU and GPU to reduce or even avoid data transmission.
  • major components such as CPU, GPU, memory, etc.
  • the CPU specifically includes: an original signal receiving module 111, configured to receive an ECG original signal, and store the ECG original signal in the first memory; the first pre-processing module 112 is configured to load the ECG original signal of the first memory to perform first-level noise reduction. Processing, obtaining a first-level ECG signal, storing the first-level ECG signal in the second memory; the first abnormal heart beat classification module 113 is configured to acquire the feature data according to the storage location information, and perform abnormality on the feature data according to the set rule determination manner The heart beat is classified, and the suspected abnormal heart beat data is obtained, and the suspected abnormal heart beat data is stored in the second memory.
  • the original signal receiving module may be connected to the electrocardiogram sensor; the electrocardiogram sensor can sense the action potential waveform of the cells in different regions of the heart, and convert the signal into an outputtable signal, which is the ECG original signal.
  • the first pre-processing module may include an IFR (Infinite Impulse Response) filter for filtering the ECG original signal to obtain a first-order ECG signal; of course, it may also be implemented by other filters, for example, FIR (Finite Impulse Response, finite-length unit impulse response) filter. Due to the tight coupling mode of the IIR filter, parallelization is difficult to implement, and therefore, the IIR filter is implemented in the CPU.
  • IFR Intelligent Impulse Response
  • FIR Finite Impulse Response, finite-length unit impulse response
  • the first abnormal heartbeat classification module may obtain a pre-defined rule determination manner from the first memory, and the rule determination manner may be implemented in the form of a parameter threshold, for example, if one or more parameters in the feature data are greater than Corresponding thresholds may preliminarily determine that the ECG signal is abnormal, and may further classify the types of abnormalities according to the threshold, obtain suspected abnormal heartbeat data, and then save the suspected abnormal heartbeat data.
  • the first abnormal heartbeat classification module processes the suspected abnormal heartbeat data, usually first stored in the first memory; due to subsequent doubts The processing of the abnormal heartbeat data reconfirmation is performed by the GPU, so the integrated memory copies the classification result to the second memory for the GPU to acquire.
  • the GPU specifically includes: a feature detecting module 121, configured to load a first-level ECG signal from the second memory for feature extraction to obtain feature data; store the feature data in the second memory; and use the second pre-processing module 122 to load the first-level
  • the ECG signal is subjected to two-stage noise reduction processing to obtain a secondary ECG signal
  • the second abnormal heart beat classification module 123 is configured to acquire a secondary ECG signal from the second pre-processing module, and obtain suspected abnormal heart beat data from the second memory, according to The set template matching method and the secondary ECG signal reconfirm the suspected abnormal heart beat data to obtain the final abnormal heart beat data, and store the final abnormal heart beat data in the second memory.
  • the above feature detection module can be implemented by various feature extraction algorithms, for example, machine learning, wavelet transform, morphological transformation, etc., in view of the particularity of the ECG signal, in order to balance the accuracy and efficiency of ECG signal feature recognition, the present embodiment
  • the feature detection module includes: (1) a morphological transformation unit, configured to transform a first-order ECG signal, and output an ECG signal in a morphological form; and (2) an R-wave detection unit, It is used for R wave detection of morphological ECG signals and outputs detection results.
  • QRS complex detection is used to detect QRS complexes of detection results and output characteristic data containing QRS complexes.
  • the R wave is the first forward wave located above the reference horizontal line in the signal period;
  • the QRS complex includes the R wave, the Q wave, the S wave, the R′ wave, the S′ wave, and the QS wave;
  • a variety of characteristic data can be obtained by parameters such as width, time interval, amplitude, and shape.
  • the feature data extracted by the feature detecting module is usually stored in the second memory; since the CPU can access the second memory, but needs to map the data stored in the second memory to the first memory; based on the above, the integrated memory includes the mapping
  • the module 103 is configured to map the feature data and the storage location information of the final abnormal heartbeat data to the first memory, so that the CPU acquires the corresponding data according to the storage location information. This method can make the CPU obtain the data processed by the GPU relatively quickly, avoiding the time consuming of the data transmission, thereby improving the analysis efficiency of the ECG signal.
  • the second pre-processing module may include an artifact removal unit configured to perform artifact removal processing on the primary ECG signal to obtain a secondary ECG signal.
  • the sensor detects different types of interference from the surface of the body surface electrode, such as power frequency interference, baseline drift, electrode contact noise, myoelectric interference, and motion interference; these disturbances form artifacts in the ECG signal, in order to A relatively pure ECG signal is obtained to improve the accuracy of subsequent feature detection and cardiac abnormality recognition.
  • the artifact removal processing is performed on the first-level ECG signal by using the above-described artifact removal unit.
  • the second abnormal heart beat classification module acquires the secondary ECG signal from the second pre-processing module on the one hand, and acquires the suspected abnormal heart beat data from the second memory on the other hand, and the suspected abnormal heart beat data is copied from the first memory to the first
  • the second abnormal heart beat classification module generates a QRS standard template through the secondary ECG signal of the noiseless signal, and then corrects the misjudged data of the suspected abnormal heart beat data according to the standard template to generate the final abnormal heart beat data.
  • the final abnormal heartbeat data is saved to the second memory, and the address of the final abnormal heartbeat data in the second memory is mapped to the first memory for the CPU to acquire. After the CPU obtains the final abnormal heartbeat data, the CPU may push the data to the user terminal, upload to the cloud platform, or perform other processing.
  • the parallel analysis device for the above ECG signals can be further optimized by the work group size, data vectorization operation and zero memory copy and copy technology to improve the analysis efficiency.
  • the CPU and the GPU jointly process various tasks in the ECG signal analysis process, and the GPU performs complex computation tasks in a parallel processing manner, thereby improving the analysis efficiency of the ECG signal, thereby improving the timeliness of the analysis feedback of the ECG signal. At the same time, it reduces the power consumption of the device and improves the user experience.
  • the method is applied to a mobile terminal, and the mobile terminal includes: an integrated memory, a CPU and a GPU, and the integrated memory includes a CPU The first memory used and the second memory for the GPU, and the CPU can access the second memory; the CPU and the GPU transfer data through the integrated memory.
  • the method comprises the following steps:
  • Step S302 the CPU performs a first-level noise reduction process on the received ECG original signal to obtain a first-level ECG signal;
  • Step S304 the GPU performs feature extraction on the first-level ECG signal to obtain feature data.
  • Step S306 the CPU performs an abnormal heartbeat classification preliminary screening process on the feature data to obtain suspected abnormal heartbeat data
  • Step S308 the GPU performs second-level noise reduction processing on the first-level ECG signal to obtain the second-level ECG signal; and applies the template matching classification method to process the suspected abnormal heart beat data and the second-level ECG signal to obtain the final abnormal heart beat data.
  • the parallel analysis method of the ECG signal provided by the embodiment of the present disclosure, the CPU performs the first-level noise reduction processing on the original ECG signal, and the GPU extracts the feature of the first-level ECG signal obtained by the first-level noise reduction processing;
  • the data is subjected to initial screening of abnormal heartbeat classification, and the GPU performs secondary noise reduction processing on the first-level ECG signal, and then applies the template matching classification method to obtain the suspected abnormal heartbeat data and the second-level noise reduction processing obtained by the primary screening process.
  • the ECG signal is processed to obtain the final abnormal heart rate data.
  • the CPU and the GPU jointly process various tasks in the ECG signal analysis process, and the GPU performs complex computation tasks in a parallel processing manner, thereby improving the analysis efficiency of the ECG signal, thereby improving the timeliness of the analysis feedback of the ECG signal. At the same time, it reduces the power consumption of the device and improves the user experience.
  • Step S402 the CPU receives the ECG original signal, and stores the ECG original signal in the first memory;
  • Step S404 the CPU loads the ECG original signal of the first memory to perform a first-level noise reduction process, to obtain a first-level ECG signal;
  • Step S406 the CPU stores the first-level ECG signal in the first memory
  • Step S408 the integrated memory copies the first-level ECG signal stored by the CPU in the first memory to the second memory;
  • Step S410 the GPU loads the first-level ECG signal from the second memory to perform feature extraction, to obtain feature data.
  • Step S412 the GPU stores the feature data in the second memory
  • Step S414 the integrated memory maps the storage location information of the feature data stored in the second memory by the GPU to the first memory
  • Step S420 the integrated memory copies the suspected abnormal heartbeat data stored by the CPU in the first memory to the second memory;
  • Step S422 the GPU loads the first-level ECG signal for the second-level noise reduction process to obtain the second-level ECG signal; in order to make full use of the heterogeneous computing resources of the CPU and the GPU, the second-level noise reduction of the ECG signal is performed in the method. Processing, from the front of the feature extraction to the front of the abnormal heartbeat reconfirmation.
  • Step S424 the GPU acquires the suspected abnormal heart beat data from the second memory, and reconfirms the suspected abnormal heart beat data according to the set template matching manner and the secondary ECG signal to obtain the final abnormal heart beat data;
  • Step S428, the integrated memory maps storage location information of the final abnormal heartbeat data stored in the second memory by the GPU to the first memory
  • Step S430 the CPU acquires the final abnormal heart rate data
  • Step S432 uploading the final abnormal heartbeat data to the remote medical platform, so that the medical platform generates a feedback report according to the final abnormal heartbeat data;
  • step S434 the CPU receives a report of the medical platform based on the feedback of the final abnormal heart rate data.
  • the CPU and the GPU jointly process various tasks in the ECG signal analysis process, and the GPU performs complex computation tasks in a parallel processing manner, thereby improving the analysis efficiency of the ECG signal, thereby improving the timeliness of the analysis feedback of the ECG signal. At the same time, it reduces the power consumption of the device and improves the user experience.
  • an embodiment of the present disclosure further provides a mobile terminal, where the mobile terminal includes the parallel analysis apparatus of the ECG signal.
  • Embodiments of the present disclosure also provide a machine readable storage medium storing machine executable instructions that, when invoked and executed by a processor, cause the processor to implement
  • machine executable instructions that, when invoked and executed by a processor, cause the processor to implement
  • a parallel analysis apparatus, method and mobile terminal for ECG signals propose a novel parallel automatic ECG analysis method using a mobile graphics processing unit (GPU); compared with the sequential analysis method of ECG signals, Reorganize the entire program flow in parallel, make full use of CPU/GPU heterogeneous computing resources, which can greatly shorten the execution time of 24-hour ECG data; optimize data in various aspects such as vectorization, workgroup size adjustment and zero-storage replication.
  • the above execution time is further shortened, feedback efficiency is improved, and the user experience is improved.
  • the average power consumption of the test mobile device is small, which alleviates the problem that the mobile device battery has limited working time.
  • each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code that includes one or more Executable instructions for logic functions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may also occur in a different order than those illustrated in the drawings.
  • each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented in a dedicated hardware-based system that performs the specified function or function. Or it can be implemented by a combination of dedicated hardware and computer instructions.
  • each functional module or unit in various embodiments of the present disclosure may be integrated to form a separate part, or each module may exist separately, or two or more modules may be integrated to form a separate part.
  • the functions may be stored in a computer readable storage medium if implemented in the form of a software functional unit and sold or used as a standalone product. Based on such understanding, the portion of the technical solution of the present disclosure that contributes in essence or to the prior art or the portion of the technical solution may be embodied in the form of a software product stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present disclosure.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Cardiology (AREA)
  • Molecular Biology (AREA)
  • Physics & Mathematics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biophysics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Signal Processing (AREA)
  • Data Mining & Analysis (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Physiology (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

一种ECG信号的并行分析装置、方法和移动终端。该装置包括集成内存(10)、CPU(11)和GPU(12)，该集成内存(10)包括供CPU(11)使用的第一内存(101)和供GPU(12)使用的第二内存(102)，且CPU(11)可访问第二内存；CPU(11)对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；还对GPU(12)提取出的特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；GPU(12)对一级ECG信号进行特征提取，得到特征数据；还对一级ECG信号进行二级降噪处理，得到二级ECG信号，应用模板匹配分类方式对疑似异常心搏数据和二级ECG信号进行处理，得到最终异常心搏数据。该装置和方法可以提高ECG信号的分析效率，从而提高ECG信号的分析反馈的及时性。

Description

ECG信号的并行分析装置、方法和移动终端
相关申请的交叉引用
本申请要求于2017年11月21日提交中国专利局的申请号为2017111696640、名称为“ECG信号的并行分析装置、方法和移动终端”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本公开涉及心电信号处理技术领域，尤其是涉及一种ECG信号的并行分析装置、方法和移动终端。
背景技术
ECG(Electrocardiogram，心电图)可以展示心脏电活动随时间的演变，是重要的生理数据之一；心率、节律的紊乱，或者心电信号的形态学变化均可能成为病理学的指征；通过分析记录的ECG波形可以检测心肌梗塞、心肌病、心肌炎等多种心脏疾病。
为了监测长期ECG信号，需要高性能服务器提供运算服务；当用户在网络不稳定的环境和海量的心电分析请求同时提交时，传统的云平台式ECG信号分析很难实时响应；如果把ECG信号分析任务转移至移动终端，由于移动终端的CPU(中央处理器，Central Processing Unit)性能有限，依然难以负担长期ECG信号的处理，并及时作出反馈；同时，处理过程需要消耗设备较大的电量，对于电池容量有限的移动终端，电池损耗较大。
发明内容
有鉴于此，本公开的目的在于提供一种ECG信号的并行分析装置、方法和移动终端，以提高ECG信号的分析效率，从而提高ECG信号的分析反馈的及时性。
为了实现上述目的，本公开采用的技术方案如下：
第一方面，本公开提供了一种ECG信号的并行分析装置，该装置包括：集成内存、CPU和GPU，集成内存包括供CPU使用的第一内存和供GPU使用第二内存，且CPU可访问第二内存；CPU和GPU通过集成内存传输数据；CPU用于对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；以及用于对GPU提取出的特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；GPU用于对一级ECG信号进行特征提取，得到特征数据；以及用于对一级ECG信号进行二级降噪处理，得到二级ECG信号，应用模板匹配分类方式对疑似异常心搏数据和二级ECG信号进行处理，得到最终异常心搏数据。
结合第一方面，本公开实施方式提供了第一方面的第一种可能的实施方式，其中，上述CPU包括原始信号接收模块，用于接收ECG原始信号，将ECG原始信号存储于第一内存；第一预处理模块，用于加载第一内存的ECG原始信号进行一级降噪处理，得到一级ECG信号，将一级ECG信号存储于第二内存；第一异常心搏分类模块，用于根据存储位置信息获取特征数据，按照设定的规则判定方式对特征数据进行异常心搏分类，得到疑似异常心搏数据，将疑似异常心搏数据存储于第二内存。
结合第一方面的第一种可能的实施方式，本公开实施方式提供了第一方面的第二种可能的实施方式，其中，上述GPU包括：特征检测模块，用于从第二内存加载一级ECG信号进行特征提取，得到特征数据；将特征数据存储于第二内存；第二预处理模块，用于加载一级ECG信号进行二级降噪处理，得到二级ECG信号；第二异常心搏分类模块，用于从第二预处理模块获取二级ECG信号，从第二内存获取疑似异常心搏数据，按照设定的 模板匹配方式和二级ECG信号对疑似异常心搏数据进行再确认，得到最终异常心搏数据，将最终异常心搏数据存储于第二内存。
结合第一方面的第二种可能的实施方式，本公开实施方式提供了第一方面的第三种可能的实施方式，其中，上述集成内存包括：映射模块，用于将特征数据以及最终异常心搏数据的存储位置信息映射至第一内存，以使CPU根据存储位置信息获取对应的数据。
结合第一方面的第二种可能的实施方式，本公开实施方式提供了第一方面的第四种可能的实施方式，其中，上述第一预处理模块包括IIR滤波器，用于对ECG原始信号进行滤波处理，得到一级ECG信号；第二预处理模块包括伪迹去除单元，用于对一级ECG信号进行伪迹去除处理，得到二级ECG信号。
结合第一方面的第二种可能的实施方式，本公开实施方式提供了第一方面的第五种可能的实施方式，其中，上述特征检测模块包括：形态学变换单元，用于对一级ECG信号进行变换，输出形态学形式的ECG信号；R波检测单元，用于对形态学形式的ECG信号进行R波检测，输出检测结果；QRS波群检测，用于对检测结果进行QRS波群检测，输出包含有QRS波群的特征数据。
第二方面，本公开提供了一种ECG信号的并行分析方法，该方法应用于移动终端，该移动终端包括：集成内存、CPU和GPU，集成内存包括供CPU使用的第一内存和供GPU使用第二内存，且CPU可访问第二内存；CPU和GPU通过集成内存传输数据；该方法包括：CPU对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；GPU对一级ECG信号进行特征提取，得到特征数据；CPU对特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；GPU对一级ECG信号进行二级降噪处理，得到二级ECG信号；以及应用模板匹配分类方式对疑似异常心搏数据和二级ECG信号进行处理，得到最终异常心搏数据。
结合第二方面，本公开实施方式提供了第二方面的第一种可能的实施方式，其中，上述方法还包括：CPU获取最终异常心搏数据，将最终异常心搏数据上传至远端的医疗平台；CPU接收医疗平台基于最终异常心搏数据反馈的报告。
结合第二方面或第二方面的第一种可能的实施方式，本公开实施方式提供了第二方面的第二种可能的实施方式，其中，上述CPU和GPU通过集成内存传输数据包括：集成内存将CPU存储于第一内存中的数据复制至第二内存，以及将GPU存储于第二内存中的数据的存储位置信息映射至第一内存。
第三方面，本公开提供了一种移动终端，该移动终端包括上述ECG信号的并行分析装置。
本公开实施方式提供的一种ECG信号的并行分析装置、方法和移动终端，其集成内存包括供CPU使用的第一内存和供GPU使用第二内存；通过CPU对ECG原始信号进行一级降噪处理，并对GPU提取出的特征数据进行异常心搏分类初筛处理；通过GPU对一级降噪处理得到的一级ECG信号进行特征提取，并对该一级ECG信号进行二级降噪处理，再应用模板匹配分类方式对初筛处理得到的疑似异常心搏数据和二级降噪处理得到的二级ECG信号进行处理，得到最终的异常心搏数据。该方式中，由CPU和GPU共同处理ECG信号分析过程中的各项任务，GPU以并行处理方式完成复杂的计算任务，提高了ECG信号的分析效率，从而提高了ECG信号的分析反馈的及时性，同时降低了设备功耗，提高了用户体验度。
进一步地，上述这种内存之间进行数据复制和映射的方式，相对于通过总线、通信线进行数据传输的方式，可以避免大量的数据传输，降低了数据从CPU端至GPU端的传输时长，进一步地提高了ECG信号的分析效率。
本公开的其他特征和优点将在随后的说明书中阐述，或者，部分特征和优点可以从说明书推知或毫无疑义地确定，或者通过实施本公开的上述技术即可得知。
为使本公开的上述目的、特征和优点能更明显易懂，下文特举较佳实施方式，并配合所附附图，作详细说明如下。
附图说明
为了更清楚地说明本公开具体实施方式或现有技术中的技术方案，下面将对具体实施方式或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图是本公开的一些实施方式，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本公开实施方式提供的一种ECG信号的并行分析装置的结构示意图；
图2为本公开实施方式提供的另一种ECG信号的并行分析装置的结构示意图；
图3为本公开实施方式提供的一种ECG信号的并行分析方法的流程图；
图4为本公开实施方式提供的另一种ECG信号的并行分析方法的流程图。
具体实施方式
为使本公开实施方式的目的、技术方案和优点更加清楚，下面将结合附图对本公开的技术方案进行清楚、完整地描述，显然，所描述的实施方式是本公开一部分实施方式，而不是全部的实施方式。基于本公开中的实 施方式，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施方式，都属于本公开保护的范围。
长期ECG可以用于辅助诊断间歇性心律失常等心脏疾病。用户可以通过穿戴式心脏监测设备采集ECG信号，并将该ECG信号发送至与该设备连接的云平台中；该云平台对ECG信号进行分析和诊断，再将诊断结果反馈至该监测设备，或者该用户的移动终端上。由于ECG信号的数据量较大，且云平台可能会连续不断地接收大量用户发送的ECG信号，因此，这种云平台处理ECG信号的方式给云端处理器造成了较大的计算负担，导致ECG信号处理反馈的及时性和可靠性得不到保证。
为了缓解上述问题，可以将ECG信号分析诊断的任务由穿戴式心脏监测设备、手机、平板电脑等移动终端完成；然而，由于移动终端的CPU性能有限，依然难以负担长期ECG信号的处理，并及时作出反馈；同时，处理过程需要消耗设备较大的电量，对于电池容量有限的移动终端，电池损耗较大。
针对上述ECG信号的分析方式反馈较慢的问题，本公开实施方式提供了一种ECG信号的并行分析装置、方法和移动终端；该技术可以应用于穿戴式心脏监测设备、手机、平板电脑等移动终端中，用于辅助诊断间歇性心律失常等心脏疾病的场景中。该技术可以采用相关的软件或硬件实现，下面通过实施方式进行描述。
参见图1所示的一种ECG信号的并行分析装置的结构示意图；该装置包括：集成内存10、CPU11和GPU(Graphic Processing Unit，图形处理器)12，该集成内存10包括供CPU11使用的第一内存101和供GPU12使用的第二内存102，且CPU11可访问第二内存102；CPU11和GPU12通过集成内存10传输数据；
CPU11用于对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；以及用于对GPU12提取出的特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；
GPU12用于对一级ECG信号进行特征提取，得到特征数据；以及用于对一级ECG信号进行二级降噪处理，得到二级ECG信号，应用模板匹配分类方式对疑似异常心搏数据和二级ECG信号进行处理，得到最终异常心搏数据。
上述CPU和GPU分别用于执行ECG信号分析中的不同任务，部分任务可以并行执行；例如，CPU进行异常心搏分类初筛处理时，GPU可以执行对一级ECG信号进行二级降噪处理。另外，由于CPU承担的任务量较大，对于计算量复杂的任务执行力较弱，而GPU为多核处理方式，可以采用并行的方式处理图像计算等计算量复杂的任务，或者具有固有并行特征的算法；根据任务的属性，对其执行的处理器进行合理分配可以提高ECG信号分析的效率；例如，上述特征提取步骤，该步骤通常需要图像识别、计算等，计算量大，将该步骤由GPU执行，可以大幅提高ECG信号分析的效率。
上述集成内存可以通过存储芯片实现，其中的第一内存和第二内存可以通过软件的形式分割；上述第一内存可以用于存储ECG原始信号、一级ECG信号和疑似异常心搏数据等数据；上述第二内存可以用于存储一级ECG信号、特征数据、疑似异常心搏数据、最终异常心搏数据等数据。
第一内存和第二内存可以通过复制和映射的方式进行数据传输；例如，集成内存可以将CPU存储于第一内存中的数据复制至第二内存，以及将GPU存储于第二内存中的数据的存储位置信息映射至第一内存。具体地，由于CPU既可以访问第一内存，也可以访问第二内存，因此，当CPU需要获取第二内存中的数据时，仅需要将该数据的在第二内存中的存储地址映射至第一内存，CPU根据映射的存储地址访问第二内存，进而获得相应的 数据；由于GPU仅可以访问第二内存，当GPU需要获取第一内存中的数据时，集成芯片需要将该数据从第一内存复制到第二内存中，再由GPU读取。
本公开实施方式提供的一种ECG信号的并行分析装置，其集成内存包括供CPU使用的第一内存和供GPU使用的第二内存；通过CPU对ECG原始信号进行一级降噪处理，并对GPU提取出的特征数据进行异常心搏分类初筛处理；通过GPU对一级降噪处理得到的一级ECG信号进行特征提取，并对该一级ECG信号进行二级降噪处理，再应用模板匹配分类方式对初筛处理得到的疑似异常心搏数据和二级降噪处理得到的二级ECG信号进行处理，得到最终的异常心搏数据。该方式中，由CPU和GPU共同处理ECG信号分析过程中的各项任务，GPU以并行处理方式完成复杂的计算任务，提高了ECG信号的分析效率，从而提高了ECG信号的分析反馈的及时性，同时降低了设备功耗，提高了用户体验度。
进一步地，上述这种内存之间进行数据复制和映射的方式，相对于通过总线、通信线进行数据传输的方式，可以避免大量的数据传输，降低了数据从CPU端至GPU端的传输时长，进一步地提高了ECG信号的分析效率。
参见图2所示的另一种ECG信号的并行分析装置的结构示意图；该装置在图1中所示装置基础上实现；该装置包括：集成内存10、CPU11和GPU12，该集成内存10包括供CPU11使用的第一内存101和供GPU12使用的第二内存102，且CPU11可访问第二内存102；CPU11和GPU12通过集成内存10传输数据。
智能手机等移动终端具有高度集成的电路，将主要组件(如CPU、GPU、内存等)组合到单个芯片中；该方式可以实现高带宽的数据传输；同时，超带宽的内存指标加速了内存和CPU/GPU之间的数据传输速度；另外， CPU和GPU内存集成在同一个芯片上，被嵌入式软件分开。在任务执行过程中，任务会发生转移，因此可以引入内存映射技术，将同一块物理内存映射到CPU和GPU的内存空间中，以减少或甚至避免数据传输。
上述CPU具体包括：原始信号接收模块111，用于接收ECG原始信号，将ECG原始信号存储于第一内存；第一预处理模块112，用于加载第一内存的ECG原始信号进行一级降噪处理，得到一级ECG信号，将一级ECG信号存储于第二内存；第一异常心搏分类模块113，用于根据存储位置信息获取特征数据，按照设定的规则判定方式对特征数据进行异常心搏分类，得到疑似异常心搏数据，将疑似异常心搏数据存储于第二内存。
上述原始信号接收模块可以与心电图传感器连接；该心电图传感器能感受心脏不同区域细胞的动作电位波形，并转换成可以输出的信号，该信号即为上述ECG原始信号。
上述第一预处理模块可以包括IIR(Infinite Impulse Response，无限的脉冲响应)滤波器，用于对ECG原始信号进行滤波处理，得到一级ECG信号；当然，还可以通过其他滤波器实现，例如，FIR(Finite Impulse Response，有限长单位冲激响应)滤波器。由于IIR滤波器的紧耦合模式，难以实现并行化，因此，IIR滤波器在CPU中实现。第一预处理模块处理得到一级ECG信号后，通常将一级ECG信号首先存储至第一内存；由于后续的特征提取处理由GPU执行，因此，集成内存再将该一级ECG信号复制至第二内存，以供GPU获取。
上述第一异常心搏分类模块可以从第一内存中获取预先制定好的规则判定方式，该规则判定方式可以以参数阈值的形式实现，例如，如果上述特征数据中的某个或者数个参数大于对应的阈值，则可初步确定该ECG信号存在异常，还可以根据阈值对异常的类型进行初步分类，得到疑似异常心搏数据，进而再将该疑似异常心搏数据进行保存。第一异常心搏分类模块处理得到疑似异常心搏数据，通常首先存储至第一内存；由于后续的疑 似异常心搏数据再确认的处理由GPU执行，因此，集成内存再将该分类结果复制至第二内存，以供GPU获取。
上述GPU具体包括：特征检测模块121，用于从第二内存加载一级ECG信号进行特征提取，得到特征数据；将特征数据存储于第二内存；第二预处理模块122，用于加载一级ECG信号进行二级降噪处理，得到二级ECG信号；第二异常心搏分类模块123，用于从第二预处理模块获取二级ECG信号，从第二内存获取疑似异常心搏数据，按照设定的模板匹配方式和二级ECG信号对疑似异常心搏数据进行再确认，得到最终异常心搏数据，将最终异常心搏数据存储于第二内存。
上述特征检测模块可以通过多种特征提取算法实现，例如，机器学习、小波变换、形态学变换等；考虑到ECG信号的特殊性，为了兼顾ECG信号特征识别的准确性和高效性，本实施方式具体采用下述方式实现：具体地，该特征检测模块包括：(1)形态学变换单元，用于对一级ECG信号进行变换，输出形态学形式的ECG信号；(2)R波检测单元，用于对形态学形式的ECG信号进行R波检测，输出检测结果；(3)QRS波群检测，用于对检测结果进行QRS波群检测，输出包含有QRS波群的特征数据。
ECG信号中，R波为信号周期内首先出现的位于参考水平线以上的正向波；QRS波群包括R波、Q波、S波、R’波、S’波和QS波；通过检测这些波形的宽度、时距、振幅、形态等参数，可以获得多种特征数据。
上述特征检测模块提取得到的特征数据通常保存在第二内存中；由于CPU可以访问第二内存，但需要将数据存储在第二内存的位置映射至第一内存；基于此，上述集成内存包括映射模块103，用于将特征数据以及最终异常心搏数据的存储位置信息映射至第一内存，以使CPU根据存储位置信息获取对应的数据。该方式可以使CPU较为快速地获得GPU处理得到的数据，避免了数据传输耗时，从而提高了ECG信号的分析效率。
上述第二预处理模块可以包括伪迹去除单元，用于对一级ECG信号进行伪迹去除处理，得到二级ECG信号。通常，传感器从体表电极检测到的信号中含有不同类型的干扰，例如，工频干扰、基线漂移、电极接触噪声、肌电干扰和运动干扰等；这些干扰在ECG信号中形成伪迹，为了获得较为纯净的ECG信号，以提高后续特征检测、心搏异常识别的准确性，本实施方式采用上述伪迹去除单元对一级ECG信号进行伪迹去除处理。
上述第二异常心搏分类模块一方面从第二预处理模块获取二级ECG信号，另一方面从第二内存获取疑似异常心搏数据，该疑似异常心搏数据预先从第一内存复制至第二内存；第二异常心搏分类模块通过无噪声信号的二级ECG信号生成QRS标准模板，再根据该标准模板纠正疑似异常心搏数据的错误判断的数据，生成最终异常心搏数据。该最终异常心搏数据保存至第二内存，并将该最终异常心搏数据在第二内存的地址映射至第一内存，以供CPU获取。该CPU获取到该最终异常心搏数据后，可以将该数据推送至用户终端、上传至云平台，或者进行其他处理。
另外，还可以通过工作组大小、数据向量化运算和零内存存复制技术进一步优化上述ECG信号的并行分析装置，提高分析效率。
上述方式中，由CPU和GPU共同处理ECG信号分析过程中的各项任务，GPU以并行处理方式完成复杂的计算任务，提高了ECG信号的分析效率，从而提高了ECG信号的分析反馈的及时性，同时降低了设备功耗，提高了用户体验度。
对应于上述装置实施方式，参见图3所示的一种ECG信号的并行分析方法的流程图；该方法应用于移动终端，该移动终端包括：集成内存、CPU和GPU，该集成内存包括供CPU使用的第一内存和供GPU使用第二内存，且CPU可访问第二内存；CPU和GPU通过集成内存传输数据。该方法包括如下步骤：
步骤S302，CPU对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；
步骤S304，GPU对一级ECG信号进行特征提取，得到特征数据；
步骤S306，CPU对特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；
步骤S308，GPU对一级ECG信号进行二级降噪处理，得到二级ECG信号；以及应用模板匹配分类方式对疑似异常心搏数据和二级ECG信号进行处理，得到最终异常心搏数据。
本公开实施方式提供的一种ECG信号的并行分析方法，CPU对ECG原始信号进行一级降噪处理，GPU对一级降噪处理得到的一级ECG信号进行特征提取；CPU对提取出的特征数据进行异常心搏分类初筛处理，GPU对上述一级ECG信号进行二级降噪处理，再应用模板匹配分类方式对初筛处理得到的疑似异常心搏数据和二级降噪处理得到的二级ECG信号进行处理，得到最终的异常心搏数据。该方式中，由CPU和GPU共同处理ECG信号分析过程中的各项任务，GPU以并行处理方式完成复杂的计算任务，提高了ECG信号的分析效率，从而提高了ECG信号的分析反馈的及时性，同时降低了设备功耗，提高了用户体验度。
参见图4所示的另一种ECG信号的并行分析方法的流程图；该方法在图3中所示方法基础上实现；该方法由移动终端的CPU、集成内存中的第一内存、第二内存和GPU多方交互实现；其中，该第二内存也可以称为显存；该方法包括如下步骤
步骤S402，CPU接收ECG原始信号，将ECG原始信号存储于第一内存；
步骤S404，CPU加载第一内存的ECG原始信号进行一级降噪处理，得到一级ECG信号；
步骤S406，CPU将一级ECG信号存储于第一内存；
步骤S408，集成内存将CPU存储于第一内存中的一级ECG信号复制至第二内存；
步骤S410，GPU从第二内存加载一级ECG信号进行特征提取，得到特征数据；
步骤S412，GPU将特征数据存储于第二内存；
步骤S414，集成内存将GPU存储于第二内存中的特征数据的存储位置信息映射至第一内存；
步骤S416，CPU根据存储位置信息获取特征数据，按照设定的规则判定方式对特征数据进行异常心搏分类，得到疑似异常心搏数据；
步骤S418，CPU将疑似异常心搏数据存储于第一内存；
步骤S420，集成内存将CPU存储于第一内存中的疑似异常心搏数据复制至第二内存；
步骤S422，GPU加载一级ECG信号进行二级降噪处理，得到二级ECG信号；为了充分利用CPU和GPU的异构计算资源，该方法中将心电信号进行伪迹去除的二级降噪处理，从特征提取前面调整至异常心搏再确认的前面。
步骤S424，GPU从第二内存获取疑似异常心搏数据，按照设定的模板匹配方式和二级ECG信号对疑似异常心搏数据进行再确认，得到最终异常心搏数据；
步骤S426，GPU将最终异常心搏数据存储于第二内存；
步骤S428，集成内存将GPU存储于第二内存中的最终异常心搏数据的存储位置信息映射至第一内存；
步骤S430，CPU获取最终异常心搏数据；
步骤S432，将最终异常心搏数据上传至远端的医疗平台，以使该医疗平台根据最终异常心搏数据生成反馈报告；
步骤S434，CPU接收医疗平台基于最终异常心搏数据反馈的报告。
上述方式中，由CPU和GPU共同处理ECG信号分析过程中的各项任务，GPU以并行处理方式完成复杂的计算任务，提高了ECG信号的分析效率，从而提高了ECG信号的分析反馈的及时性，同时降低了设备功耗，提高了用户体验度。
对应于上述装置和方法实施方式，本公开实施方式还提供了一种移动终端，该移动终端包括上述ECG信号的并行分析装置。
本公开实施方式还提供了一种机器可读存储介质，该机器可读存储介质存储有机器可执行指令，该机器可执行指令在被处理器调用和执行时，机器可执行指令促使处理器实现上述ECG信号的并行方法，具体实现可参见方法实施方式，在此不再赘述。
本公开实施方式提供的一种ECG信号的并行分析装置、方法和移动终端，提出了一种利用移动图形处理单元(GPU)的新型并行自动ECG分析方式；与ECG信号的顺序分析方式相比，并行方式重组整个程序流，充分利用CPU/GPU异构计算资源，该方式可以大幅缩短24小时长ECG数据的执行时间；通过数据矢量化，工作组大小调整和零存储复制等各个方面的优化，上述执行时间进一步缩短，反馈效率提升，提高了用户体验。此外，当将大量计算分配给GPU时，测试移动设备的平均功耗较小，缓解了移动设备电池工作时间有限的问题。
在本申请所提供的几个实施方式中，应该理解到，所揭露的装置和方法，也可以通过其它的方式实现。以上所描述的装置实施方式仅仅是示意性的，例如，附图中的流程图和框图显示了根据本公开的多个实施方式的装置、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段或代码的一部分，所述模块、程序段或代码的一部分包含一个或多个用于实现规定的 逻辑功能的可执行指令。也应当注意，在有些作为替换的实现方式中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个连续的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或动作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
另外，在本公开各个实施方式中的各功能模块或单元可以集成在一起形成一个独立的部分，也可以是各个模块单独存在，也可以两个或两个以上模块集成形成一个独立的部分。
所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本公开的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本公开各个实施方式所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是：以上所述实施方式，仅为本公开的具体实施方式，用以说明本公开的技术方案，而非对其限制，本公开的保护范围并不局限于此，尽管参照前述实施方式对本公开进行了详细的说明，本领域的普通技术人员应当理解：任何熟悉本技术领域的技术人员在本公开揭露的技术范围内，其依然可以对前述实施方式所记载的技术方案进行修改或可轻易想到变化，或者对其中部分技术特征进行等同替换；而这些修改、变化或者替换，并不使相应技术方案的本质脱离本公开实施方式技术方案的精神 和范围，都应涵盖在本公开的保护范围之内。因此，本公开的保护范围应所述以权利要求的保护范围为准。

Claims (10)

  1. 一种ECG信号的并行分析装置，其特征在于，所述装置包括：集成内存、CPU和GPU，所述集成内存包括供CPU使用的第一内存和供所述GPU使用的第二内存，且所述CPU可访问所述第二内存；所述CPU和所述GPU通过所述集成内存传输数据；
    所述CPU配置成对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；以及配置成对所述GPU提取出的特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；
    所述GPU配置成对所述一级ECG信号进行特征提取，得到所述特征数据；以及配置成对所述一级ECG信号进行二级降噪处理，得到二级ECG信号，应用模板匹配分类方式对所述疑似异常心搏数据和所述二级ECG信号进行处理，得到最终异常心搏数据。
  2. 根据权利要求1所述的装置，其特征在于，所述CPU包括：
    原始信号接收模块，配置成接收ECG原始信号，将所述ECG原始信号存储于所述第一内存；
    第一预处理模块，配置成加载所述第一内存的ECG原始信号进行一级降噪处理，得到一级ECG信号，将所述一级ECG信号存储于所述第二内存；
    第一异常心搏分类模块，配置成根据存储位置信息获取所述特征数据，按照设定的规则判定方式对所述特征数据进行异常心搏分类，得到疑似异常心搏数据，将所述疑似异常心搏数据存储于所述第二内存。
  3. 根据权利要求2所述的装置，其特征在于，所述GPU包括：
    特征检测模块，配置成从所述第二内存加载所述一级ECG信号进行特征提取，得到特征数据；将所述特征数据存储于所述第二内存；
    第二预处理模块，配置成加载所述一级ECG信号进行二级降噪处理，得到二级ECG信号；
    第二异常心搏分类模块，配置成从所述第二预处理模块获取所述二级ECG信号，从所述第二内存获取所述疑似异常心搏数据，按照设定的模板匹配方式和所述二级ECG信号对所述疑似异常心搏数据进行再确认，得到最终异常心搏数据，将最终异常心搏数据存储于所述第二内存。
  4. 根据权利要求3所述的装置，其特征在于，所述集成内存包括：映射模块，配置成将所述特征数据以及最终异常心搏数据的存储位置信息映射至所述第一内存，以使所述CPU根据所述存储位置信息获取对应的数据。
  5. 根据权利要求3所述的装置，其特征在于，所述第一预处理模块包括IIR滤波器，配置成对ECG原始信号进行滤波处理，得到一级ECG信号；
    所述第二预处理模块包括伪迹去除单元，配置成对所述一级ECG信号进行伪迹去除处理，得到二级ECG信号。
  6. 根据权利要求3所述的装置，其特征在于，所述特征检测模块包括：
    形态学变换单元，配置成对所述一级ECG信号进行变换，输出形态学形式的ECG信号；
    R波检测单元，配置成对所述形态学形式的ECG信号进行R波检测，输出检测结果；
    QRS波群检测，配置成对所述检测结果进行QRS波群检测，输出包含有QRS波群的特征数据。
  7. 一种ECG信号的并行分析方法，其特征在于，所述方法应用于移动终端，所述移动终端包括：集成内存、CPU和GPU，所述集成内存包括供CPU使用的第一内存和供所述GPU使用第二内存，且所述CPU可访问所述第二内存；所述CPU和所述GPU通过所述集成内存传输数据；
    所述方法包括：
    所述CPU对接收到的ECG原始信号进行一级降噪处理，得到一级ECG信号；
    所述GPU对所述一级ECG信号进行特征提取，得到特征数据；
    所述CPU对所述特征数据进行异常心搏分类初筛处理，得到疑似异常心搏数据；
    所述GPU对所述一级ECG信号进行二级降噪处理，得到二级ECG信号；以及应用模板匹配分类方式对所述疑似异常心搏数据和所述二级ECG信号进行处理，得到最终异常心搏数据。
  8. 根据权利要求7所述的方法，其特征在于，所述方法还包括：
    所述CPU获取所述最终异常心搏数据，将所述最终异常心搏数据上传至远端的医疗平台；
    所述CPU接收所述医疗平台基于所述最终异常心搏数据反馈的报告。
  9. 根据权利要求7或8所述的方法，其特征在于，所述CPU和所述GPU通过所述集成内存传输数据包括：
    所述集成内存将所述CPU存储于第一内存中的数据复制至所述第二内存，以及将所述GPU存储于所述第二内存中的数据的存储位置信息映射至所述第一内存。
  10. 一种移动终端，其特征在于，所述移动终端包括权利要求1-7任一项所述的装置。
PCT/CN2017/113408 2017-11-21 2017-11-28 Ecg信号的并行分析装置、方法和移动终端 Ceased WO2019100417A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US15/739,719 US10743789B2 (en) 2017-11-21 2017-11-28 ECG signal parallel analysis apparatus, method and mobile terminal
EP17933117.8A EP3716134B1 (en) 2017-11-21 2017-11-28 Parallel analysis device for ecg signals, and method and mobile terminal

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711169664.0 2017-11-21
CN201711169664.0A CN107958214A (zh) 2017-11-21 2017-11-21 Ecg信号的并行分析装置、方法和移动终端

Publications (1)

Publication Number Publication Date
WO2019100417A1 true WO2019100417A1 (zh) 2019-05-31

Family

ID=61965216

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/113408 Ceased WO2019100417A1 (zh) 2017-11-21 2017-11-28 Ecg信号的并行分析装置、方法和移动终端

Country Status (5)

Country Link
US (1) US10743789B2 (zh)
EP (1) EP3716134B1 (zh)
JP (1) JP6583892B2 (zh)
CN (1) CN107958214A (zh)
WO (1) WO2019100417A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114587375A (zh) * 2022-03-28 2022-06-07 联通(广东)产业互联网有限公司 心电图的预处理方法、关键波段提取方法、设备和介质
US20230190203A1 (en) * 2020-04-09 2023-06-22 Obshchestvo S Ogranichennoj Otvetstvennostyu "Parma-Telekom" Human health risk assessment method

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105232032A (zh) * 2015-11-05 2016-01-13 福州大学 基于小波分析远程心电监护与预警系统及方法
CN109171706B (zh) * 2018-09-30 2020-12-29 南京信息工程大学 基于分类匹配和分数阶扩散的心电信号去噪方法和系统
CN109730670B (zh) * 2018-10-12 2021-11-30 浙江大学宁波理工学院 一种基于异构计算的心电信号降噪方法
CN110537909A (zh) * 2019-09-03 2019-12-06 深圳旭宏医疗科技有限公司 基于心电监测设备的急救方法、装置和计算机设备
CN111161874A (zh) * 2019-12-23 2020-05-15 乐普（北京）医疗器械股份有限公司 一种智能心电分析装置
CN113040778B (zh) * 2019-12-26 2022-07-29 华为技术有限公司 诊断报告生成方法、装置、终端设备及可读存储介质
US20250228457A1 (en) * 2021-10-27 2025-07-17 Monovo, LLC Monitoring system and method for remote monitoring of physiological health
CN115299890B (zh) * 2022-07-26 2024-11-29 湖北智奥物联网科技有限公司 孕期健康管理的柔性装置及系统
CN116509360B (zh) * 2023-06-25 2023-09-12 苏州维伟思医疗科技有限公司 心律监控系统、方法及医疗设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060056730A1 (en) * 2004-08-24 2006-03-16 Ziosoft Inc. Method, computer program product, and apparatus for performing rendering
CN105266849A (zh) * 2014-07-09 2016-01-27 无锡祥生医学影像有限责任公司 实时超声弹性成像方法和系统
US20160125565A1 (en) * 2014-11-04 2016-05-05 Kabushiki Kaisha Toshiba Asynchronous method and apparatus to support real-time processing and data movement
CN105899268A (zh) * 2015-06-23 2016-08-24 中国科学院深圳先进技术研究院 基于gpu的并行心电信号分析方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8531471B2 (en) * 2008-11-13 2013-09-10 Intel Corporation Shared virtual memory
US8909332B2 (en) * 2010-01-26 2014-12-09 Stmicroelectronics S.R.L. Method and device for estimating morphological features of heart beats
US9298769B1 (en) * 2014-09-05 2016-03-29 Futurewei Technologies, Inc. Method and apparatus to facilitate discrete-device accelertaion of queries on structured data
CN106725411B (zh) * 2017-02-15 2020-10-27 成都皓图智能科技有限责任公司 一种基于gpu加速的非接触式心跳检测方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060056730A1 (en) * 2004-08-24 2006-03-16 Ziosoft Inc. Method, computer program product, and apparatus for performing rendering
CN105266849A (zh) * 2014-07-09 2016-01-27 无锡祥生医学影像有限责任公司 实时超声弹性成像方法和系统
US20160125565A1 (en) * 2014-11-04 2016-05-05 Kabushiki Kaisha Toshiba Asynchronous method and apparatus to support real-time processing and data movement
CN105899268A (zh) * 2015-06-23 2016-08-24 中国科学院深圳先进技术研究院 基于gpu的并行心电信号分析方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
SALVATORE CUOMO: "A GPU-parallel algorithm for ECG signal denoising based on the NLM method", IEEE COMPUTER SOCIETY; 2016 30TH INTERNATIONAL CONFERENCE ON ADVANCED INFORMATION NETWORKING AND APPLICATIONS WORKSHOPS, 25 March 2016 (2016-03-25), pages 35 - 39, XP032902390, DOI: 10.1109/WAINA.2016.110 *
See also references of EP3716134A4 *
WENFENG SHEN: "Load-prediction scheduling for Computer Simulation of Ele- ctrocardiogram on a CPU-GPU PC", PROCEEDINGS OF THE 2013 IEEE 16TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND ENGINEERING, 31 December 2013 (2013-12-31), XP032573359, DOI: 10.1109/CSE.2013.42 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20230190203A1 (en) * 2020-04-09 2023-06-22 Obshchestvo S Ogranichennoj Otvetstvennostyu "Parma-Telekom" Human health risk assessment method
CN114587375A (zh) * 2022-03-28 2022-06-07 联通(广东)产业互联网有限公司 心电图的预处理方法、关键波段提取方法、设备和介质

Also Published As

Publication number Publication date
US10743789B2 (en) 2020-08-18
EP3716134A1 (en) 2020-09-30
JP2019093096A (ja) 2019-06-20
CN107958214A (zh) 2018-04-24
EP3716134B1 (en) 2023-01-04
EP3716134A4 (en) 2021-08-04
US20190150778A1 (en) 2019-05-23
JP6583892B2 (ja) 2019-10-02

Similar Documents

Publication Publication Date Title
WO2019100417A1 (zh) Ecg信号的并行分析装置、方法和移动终端
CN107122788B (zh) 一种基于心电信号的身份识别方法及装置
CN109165556B (zh) 一种基于grnn身份识别方法
JP6335199B2 (ja) ハンドヘルド・デバイスを用いた非侵襲的な脈波伝播時間の判定
WO2021071688A1 (en) Systems and methods for reduced lead electrocardiogram diagnosis using deep neural networks and rule-based systems
CN108836307A (zh) 一种心电智能检测装置、设备及移动终端
CN119993462B (zh) 多模块协同的快速心电图自动诊断方法及系统
Wang et al. Wound image analysis system for diabetics
CN109077720B (zh) 信号处理方法、装置、设备和存储介质
Torti et al. Custom FPGA processing for real-time fetal ECG extraction and identification
CN114469132A (zh) 模型训练方法、装置、电子设备及存储介质
CN110200620A (zh) 多模态心脏数据检测方法、装置、存储介质及终端设备
CN114587373B (zh) 一种基于硬件部署的心电采集、识别、存储一体化系统
CN116644353A (zh) 非接触式心率测量方法和装置
WO2021031155A1 (zh) 一种基于ecg的多尺度特征提取方法及装置
CN120543944A (zh) 一种基于图像分析的医学症状确定方法及系统
CN119344746A (zh) 基于人工智能的心电异常数据分析方法及系统
CN119732660A (zh) 血流动力学参数的检测方法、终端设备和系统
JP2025022843A (ja) 心電図の視覚表現の改善のための装置及び方法
Shu et al. Image-decomposition-enhanced deep learning for detection of rotor cores in cardiac fibrillation
CN115691799A (zh) 猝死风险预测模型的训练方法、猝死风险预测方法和装置
CN109730670B (zh) 一种基于异构计算的心电信号降噪方法
CN115227256B (zh) 一种基于向量环的心电信号压缩方法、终端设备及介质
Pendke et al. ECG VisionSense: A Cloud-Deployed, Flask-Based Intelligent Cardiovascular Disease Diagnosis System
Ragulskis Clustering of ECG segments for patients before sudden cardiac death based on Lagrange descriptors

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17933117

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017933117

Country of ref document: EP

Effective date: 20200622