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
  • 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
  • A61B5/72—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
  • A61B5/7207—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal of noise induced by motion artifacts
• • 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]
  • A61B5/346—Analysis of electrocardiograms
  • A61B5/349—Detecting specific parameters of the electrocardiograph cycle

Definitions

• the present invention relates to a motion artifact removal technique for processing physiological signals and more particularly to a method and apparatus for reducing motion artifacts in ECG signals.
• Bio signals for example electrocardiogram (ECG) signals
• ECG signals may include substantial amounts of noise.
• noise introduced by muscular activity, motion artifacts, etc, which usually appear as rapid, wavy deflections that render the ECG difficult to read, especially during the movement.
• US 6216031B 1 proposes an apparatus for enhancing signals in an ECG including artifacts, which apparatus comprises a mean value unit for evaluating the curve shape of a predetermined number of ECG signal beats from the beginning of a QRS complex to the end of a T wave, and using the outcome to establish a mean value beat, and it further comprises a subtracting unit for subtracting the ECG signal's mean value beat from an actual beat to obtain a residual signal, and a FIR filter unit for high-low-pass filtering the residual signal to provide a filtered signal to which the mean value beat is added in an adding unit.
• US 621603 IB 1 The principle of US 621603 IB 1 is to utilize the mean value unit to calculate the mean value beat based on averaging of successive ECG cycles and utilize the LP (low pass) and HP (high pass) filters to deal with residual signals to remove both muscle noises and baseline wander noises. Although this method enhances the signal-to-noise ratio (SNR) of the resulted ECG signal to some extent, it still has some unacceptable defects.
• SNR signal-to-noise ratio
• the residual signal is filtered by a low-pass filter to reduce muscle noise, and by a high-pass filter to reduce baseline wander.
• the cutoff frequencies of the filters are set to values that avoid unacceptable distortion of the remaining P waves and VPBs in the residual signal.
• Both filters are finite impulse response filters with the advantage that the delay of the filtered residual signal is constant and signal-independent. But unacceptable notches may also appear, particularly in the case of an abrupt ECG morphology change.
• a method of reducing motion artifacts in ECG signals comprising: acquiring a current beat from a continuously measured ECG signal of a patient; calculating a correlation coefficient between a previous mean value beat and the current beat in the ECG signal; determining the weights to be assigned to the previous mean value beat and the current beat based on the correlation coefficient; and calculating a current mean value beat based on the previous mean value beat, the current beat, and the weights thereof.
• weight determination is based on correlation calculation
• different weights can be dynamically assigned to both the previous mean value beat and the current beat according to the correlation between current beat waveform and previous mean value beat.
• the correlation is high, a larger weight will be assigned to the previous mean value beat. Otherwise, it will be assigned a smaller weight, even a zero-value weight, which implies that the current beat is not correlated with the mean beat and not used for the mean beat update.
• the method in accordance with the present invention adopts an adaptive weight determination method, which is based on the correlation between the two beats and which may grasp the exact ECG morphology change, resulting in an enhanced filtering effect such that the artifacts due to motion of the patient can be removed.
• an adaptive weight determination method which is based on the correlation between the two beats and which may grasp the exact ECG morphology change, resulting in an enhanced filtering effect such that the artifacts due to motion of the patient can be removed.
• a mean value beat should be calculated or empirically determined as the initial mean value beat before the start of the real filtering of the ECG signal.
• the initial mean value beat will be used as the previous mean value beat and will be combined with the first current beat to calculate the first current mean value beat, which will be stored in a memory. Then, the first current mean value beat will be subtracted from the first current beat to obtain a residual signal if these two beats are correlated, and the residual signal will be filtered to obtain a filtered signal. Finally, the filtered signal will be added back to the first current mean value beat to obtain a modified first current beat.
• the stored first current mean value beat will be used as the previous mean value beat to be combined with the second current beat to calculate the second current mean value beat, which will be stored in a memory. Then, if the second current beat is correlated with the previous mean beat, the second current mean value beat will be subtracted from the second current beat to obtain a residual signal, and the residual signal will be filtered to obtain a filtered signal. Finally, the filtered signal will be added back to the second current mean value beat to obtain a modified second current beat. The processing operation continues like this and finally the ECG signals are all processed and filtered.
• the correlation between the previous mean value beat and the current beat may be a correlation coefficient, but is not limited thereto, as will be easily understood by the person skilled in the art.
• the weights to be assigned to the previous mean value beat and the current beat may be determined empirically or by mapping with a lookup table.
• a preferred example of this method may further comprise a step of comparing the correlation coefficient with a predetermined coefficient, i.e., a predetermined correlation threshold. If the correlation coefficient is higher than the predetermined coefficient, a bigger weight will be assigned to the previous mean value beat. Otherwise, the previous mean value beat will be assigned a smaller weight or even a zero-value weight.
• the method may further comprise the steps of: subtracting the current mean value beat from the current beat to obtain a residual signal; performing piecewise filtering on the residual signal based on signal characteristics of the ECG signal to obtain a filtered signal; and adding the current mean value beat back to the filtered signal to obtain a modified current beat.
• the signal characteristics may be acquired from the ECG signal and may include the peak positions of P wave, QRS complex wave, and T wave, the segment intervals, and the noise envelope and the correlation information with the mean beat.
• the step of performing piecewise filtering may further comprise the sub-steps of: identifying the QRS segment and T and P segments in the ECG wave based on the signal characteristics, estimating the noise envelope based on the signal slope change, and the calculated correlation coefficient; and performing filtering on the different segments, respectively, by using different filters.
• the method according to the preferred example of the present invention adopts piecewise filtering of the residual signal to improve the effect of the noise reduction to enhance the S R and to guarantee that the important features are undistorted.
• the term "piecewise filtering" means that, for example, if a QRS-complex wave is selected as the beat to be processed, one group of filters is used for one QRS segment and another group of filters is used for T and P segments based on the current beat signal characteristics, peak locations and interval information of the ECG morphology.
• the subtraction of the current mean value beat from the current beat to obtain a residual signal is applied in the case when the current beat is correlated with the mean beat. Otherwise, the current beat can be reserved, for example, in the residual signal, and a different group of piece- wise filtering can be applied to keep the ECG features exact. In this situation, the group of piece-wise filters used for the uncorrected beat is totally different from that used for the correlated beat.
• the method may further comprise a step of deriving an initial mean value beat from a predetermined number of consecutive beats in the ECG signal. Since the initial mean value beat is derived from the measurement performed on the patient himself, the accuracy of the mean value beat thus obtained may be improved.
• the predetermined number of consecutive beats comprises a first predetermined number of consecutive beats and a second predetermined number of consecutive beats following the first predetermined number of consecutive beats.
• the step of deriving an initial mean value beat further comprises the sub-steps of deriving an initial mean value beat from the first predetermined number of consecutive beats; and updating the initial mean value beat only when the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is larger than a predetermined coefficient.
• the initial mean value beat is carefully validated by introducing a step of updating the initial mean value beat only if the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is larger than a predetermined coefficient.
• the conventional way of deriving an initial mean value beat from a predetermined number of consecutive beats only relates to deriving an initial mean value beat from the first predetermined number of consecutive beats by averaging or some arithmetic algorithms.
• the method according to the present invention further utilizes a second predetermined number of consecutive beats to validate and update the initial mean value beat.
• the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is also used to update the initial mean value beat.
• the initial mean value beat may be derived more accurately, and accordingly the accuracy of the mean value beat thus obtained may be improved and the final filtered ECG signal may be more accurate and the SNR of the ECG signal can be further improved.
• an apparatus for reducing motion artifacts in ECG signals comprising: an acquiring unit for acquiring a current beat from a continuously measured ECG signal of a patient; a first calculating unit for calculating a correlation coefficient between a previous mean value beat and the current beat in the ECG signal; a determining unit for determining the weights to be assigned to the previous mean value beat and the current beat based on the correlation coefficient; and a second calculating unit for calculating a current mean value beat based on the previous mean value beat, the current beat, and the weights thereof.
• the apparatus in accordance with the present invention adopts an adaptive weight
• Fig. 1 is a flow chart of the method according to the invention.
• Fig. 2 shows a flow chart of a variation of the method according to the invention
• FIG. 3 shows a flow chart of another variation of the method according to the invention
• Fig. 4 shows a flow chart of still another variation of the method according to the invention
• Fig. 5 is a block diagram of the apparatus according to the invention.
• Fig. 6 shows how motion artifacts are removed from an ECG signal by utilizing the method of the invention.
• Fig. 1 is a flow chart of the method 10 of reducing motion artifacts in ECG signals in accordance with one embodiment of the present invention.
• the method 10 as shown in Fig. 1 is directed to an improved method of calculating the current mean value beat to be used for the filtering of the current beat.
• it is characterized in that it adopts an adaptive and dynamic determination of the weights to be assigned to the previous mean value beat and the current beat based on the correlation coefficient therebetween, and then the calculation of a current mean value beat is performed based on the previous mean value beat, the current beat, and the weights thereof.
• Fig. 5 is a block diagram of the apparatus 50 for implementing the method 10 shown in Fig. 1.
• the apparatus 50 for reducing motion artifacts in ECG signals in accordance with one embodiment of the present invention comprises an acquiring unit 51, a first calculating unit 52, a determining unit 53, and a second calculating unit 54.
• an ECG sensor on a patient may continuously measure the ECG signal of the patient and the obtained ECG signal will be input to the acquiring unit 51, which may acquire a current beat from the received ECG signal (step 11 in Fig. 1).
• the first calculating unit 52 is coupled with the acquiring unit 51 and is capable of calculating a correlation coefficient between a previous mean value beat and the current beat in the ECG signal (step 12 in Fig. 1).
• a mean value beat should be calculated or empirically determined as the initial mean value beat before the start of the real filtering of the ECG signal.
• the initial mean value beat will be used as the previous mean value beat and will be combined with the first current beat to calculate the first current mean value beat, which will be stored in a memory. Then, the first current mean value beat will be subtracted from the first current beat to obtain a residual signal if these two beats are correlated, and the residual signal will be filtered to obtain a filtered signal. Finally, the filtered signal will be added back to the first current mean value beat to obtain a modified first current beat.
• the stored first current mean value beat will be used as the previous mean value beat to be combined with the second current beat to calculate the second current mean value beat, which will be stored in a memory. Then, if the second current beat is correlated with the mean beat, the second current mean value beat will be subtracted from the second current beat to obtain a residual signal, and the residual signal will be filtered to obtain a filtered signal. Finally, the filtered signal will be added back to the second current mean value beat to obtain a modified second current beat. The processing operation continues in this manner until finally the ECG signals are all processed and filtered.
• the subtraction and addition of the mean beat from and to the current beat are applied in the situation where the current beat and the mean beat are correlated. Otherwise, the current beat may be reserved, for example in the residual signal, and then filtered to generate the output from which the noise has been removed.
• the correlation between the previous mean value beat and the current beat may be a correlation coefficient, but is not limited thereto.
• the correlation may be also the mean-squared value of the amplitude error or both, etc, as will be easily understood by those skilled in the art.
• the determining unit 53 is coupled with the first calculating unit 52 and may determine the weights to be assigned to the previous mean value beat and the current beat based on the correlation coefficient (step 13 in Fig. 1).
• the weights to be assigned to the previous mean value beat and the current beat may be determined empirically or by mapping with a lookup table.
• the second calculating unit 54 which is coupled with the determining unit 53, is utilized for calculating a current mean value beat based on the previous mean value beat, the current beat, and the weights assigned to these beats (step 14 in Fig. 1).
• weights can be dynamically assigned to both the previous mean value beat and the current beat according to the correlation between current beat waveform and previous mean value beat. In other words, if the correlation is high, a larger weight will be assigned to the previous mean value beat. Otherwise, it will be assigned a smaller weight or even a zero- value weight.
• the method 10 in accordance with the present invention adopts an adaptive weight determination method, which is based on the correlation between the two beats and which may result in good tracking of the ECG morphology change and in a better filtering effect, such that the artifacts due to patient motion can be removed.
• Figs. 2-4 show some possible variations to the method 10 of reducing motion artifacts in ECG signals in accordance with the present invention.
• the method may further comprise a step 15 of comparing the correlation coefficient with a predetermined coefficient, i.e., a predetermined correlation threshold.
• a correlation coefficient CCR is calculated by the correlation calculation unit and compared with a pre-determined correlation threshold, indicating whether the current cycle is highly correlated with the existing mean cycle: highly correlated
• the correlation coefficient is higher than the predetermined coefficient, a larger weight will be assigned to the previous mean value beat. Otherwise the previous mean value beat will be assigned a smaller weight or even a zero-value weight.
• Fig. 3 shows a flow chart of the method 10 of reducing motion artifacts in ECG signals in accordance with a further embodiment of the present invention.
• the method may further comprise a step 16 of subtracting the current mean value beat from the current beat to obtain a residual signal, after the current mean value beat has been calculated based on the previous mean value beat, the current beat, and the weights thereof.
• step 16 of subtraction is not applied and the beat is reserved, for example, in the residual signal directly.
• the method 10 further comprises a step 17 of performing piecewise filtering on the residual signal based on signal characteristics of the ECG signal to obtain a filtered signal.
• the method may further comprise a step 18 in which the current mean value beat is added back to the filtered signal to obtain a modified current beat when the beat is correlated with the mean beat. Otherwise, said adding step is omitted.
• the current mean value beat obtained in the method 10 shown in Fig. 1 is utilized in the real filtering of the ECG signal to finally achieve that motion artifacts are removed from the ECG signal.
• the signal characteristics may be acquired from the ECG signal and may include the peak positions of P wave, QRS complex wave, and T wave, and the segment intervals.
• step 17 of performing piecewise filtering may further comprise the sub-steps of identifying the QRS segment and T and P wave segments, the noise envelope based on the signal characteristics, and the correlation with the mean beat; and performing filtering on the different segments, respectively, by using different filters.
• the method 10 according to the preferred example of the present invention shown in Fig. 3 adopts piecewise filtering of the residual signal to enhance the ECG's SNR and to make sure that the ECG features remain undi started in the motion artifact-removed signal.
• the term "piecewise filtering" means that, for example, if a QRS complex wave is selected as the beat to be processed, one group of LP and HP filters is used for one QRS segment and another group of LP and HP filters is used for T and P segments based on the current beat signal characteristics i.e., peak locations, interval information, the noise envelopment of the ECG morphology and the correlation information with the mean beat.
• the first group of low-pass filters and high-pass filters for the QRS segment and a second group of low-pass filters and high-pass filters for other segments may be different in cutoff frequency.
• the two groups of filters may be different in other parameters, as long as they are specifically selected in accordance with the signal
• piecewise LP/HP filtering based on beat characteristics can be used to deal with the notches due to the abrupt change between QRS transitions, and accordingly, the SNR of the ECG signal can be improved substantially. Moreover, at the same time, important ECG features can be perfectly maintained.
• a constant delay of is introduced to the piecewise filtered signal; therefore, in order to obtain the correct artifact- removed signal, when adding the mean value beats back to the filtered signal, the constant delay should be considered to compensate for the positions of the adding operations
• a different type of HP filter may be utilized, for example, a butterworth filter with a flat pass-band.
• Fig. 4 shows a flow chart of the method 10 of reducing motion artifacts in ECG signals in accordance with still another embodiment of the present invention.
• the only difference between the methods 10 shown in Fig. 3 and Fig. 4 is that in Fig. 4 it further comprises a step 19 of acquiring, from the ECG signal, the signal characteristics, and a step 20 of selecting one of P wave, QRS complex wave, and T wave, or any combination thereof as the beat to be processed.
• a raw ECG signal is processed to get signal characteristics, such as QRS positions and segment intervals, as mentioned above.
• the related signal characteristics may be sent to a current signal template unit for classifying current signal templates.
• a different scheme of multiple signal templates (QRS template, T template and QRST template) may be utilized to improve the tracing of the motion artifact-removed ECG signals, which results in more accurate feature extraction for ECG diagnosis.
• the templates have predetermined lengths, with the central and most visually obvious parts of the tracing as the reference portions for alignment in these templates.
• ECG signals vary, the proposed multiple templates, and the previously discussed dynamic determined weights, ensure that the ECG signal is perfectly tracked by the updated mean value beat templates.
• a mean value beat should be calculated or empirically determined as the initial mean value beat before the start of the real filtering of the ECG signal.
• the calculation of the mean value beat is a key factor and may have great influence on the final ECG signal obtained, the calculation of the initial mean value beat is particularly important.
• the initial mean value beat may be determined empirically by referring to a normal beat of ordinary people.
• the method may further comprise a step of deriving an initial mean value beat from a predetermined number of consecutive beats in the ECG signal. Since the initial mean value beat is derived from a measurement on the patient himself, the accuracy of the mean value beat thus obtained may be improved.
• the predetermined number of consecutive beats may comprise two portions, i.e., a first predetermined number of consecutive beats and a second predetermined number of consecutive beats following the first predetermined number of consecutive beats.
• the step of deriving an initial mean value beat may further comprise the sub-steps of deriving an initial mean value beat from the first predetermined number of consecutive beats; and updating the initial mean value beat only when the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is larger than a predetermined coefficient.
• the initial mean value beat is carefully validated by introducing a step of updating the initial mean value beat only when the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is larger than a predetermined coefficient
• the conventional way of deriving an initial mean value beat from a predetermined number of consecutive beats only relates to deriving an initial mean value beat from the first predetermined number of consecutive beats by averaging or some arithmetic algorithms.
• the method 10 according to the present invention further utilizes a second predetermined number of consecutive beats to validate and update the initial mean value beat.
• the correlation coefficient between a previous initial mean value beat and a current beat in the second predetermined number of consecutive beats is also used to update the initial mean value beat.
• the initial mean value beat may be derived more accurately, and accordingly the accuracy of the mean value beat thus obtained may be improved and the final filtered
• ECG signal may be more accurate and the SNR of the ECG signal can be further improved.
• the method may further comprise, if necessary, a step of reactivating the step of deriving an initial mean value beat from a predetermined number of consecutive beats in the ECG signal.
• the user of the apparatus 50 may operate a button or the like to reactivate the step of deriving an initial mean value beat, so that a new filtering step of the ECG signal will start. In this way, the filtered ECG signal will be timely corrected.
• Fig. 6 shows how motion artifacts are removed from ECG signals by utilizing the method of the invention.
• S I represents the original or raw ECG signal received from the patient
• S2 represents a residual signal obtained by subtracting the current mean value beat from the current beat
• S3 represents the filtered residual signal which has been subject to piecewise filtering based on signal characteristics of the ECG signal
• S4 represents the modified current beat (i.e. the motion artifact-removed signal) by adding the current mean value beat back to the filtered signal.
• the period T shown in Fig. 6 stands for the period during which the initial mean value beat is derived from a predetermined number of consecutive beats in the ECG signal
• Fig. 5 only shows a basic block diagram of the apparatus 50 according to the present invention, it may be easily understood by those skilled in the art that, there could be a unit corresponding to each step in the above described methods 10, including the steps shown in Figs. 2-4, to perform the relevant method step.
• the apparatus 50 according to the present invention may further comprise a comparator for comparing the correlation coefficient with a predetermined coefficient.
• a comparator for comparing the correlation coefficient with a predetermined coefficient.
• the apparatus 50 may further comprise a subtracting unit for subtracting the current mean value beat from the current beat to obtain a residual signal; a filter unit for performing piecewise filtering on the residual signal based on signal characteristics of the ECG signal to obtain a filtered signal; and an adding unit for adding the filtered signal back to the current mean value beat to obtain a modified current beat.
• the apparatus 50 per se may be a personal computer with a CPU and a memory, a Single-chip Microcomputer or a CPU (i.e., a processing unit) alone. Therefore, the respective units comprised therein may be implemented as software or computer-readable instructions. However, as will be easily understood by those skilled in the art, the respective units may be hardware entities as well. In other words, the apparatus 50 may be composed of distinct hardware modules. Each of the units may be implemented by a single processor or a plurality of processors.
• any reference signs placed between parentheses shall not be construed as limiting the claim.
• the word “comprising” does not exclude the presence of elements or steps not listed in a claim or in the description.
• the word “a” or “an” preceding an element does not exclude the presence of a plurality of such elements.
• several of these units can be embodied by one and the same item of software and/or hardware.
• the usage of the words first, second and third, et cetera does not indicate any ordering. These words are to be interpreted as names.

Landscapes

• Health & Medical Sciences (AREA)
• Life Sciences & Earth Sciences (AREA)
• Engineering & Computer Science (AREA)
• Cardiology (AREA)
• Public Health (AREA)
• Medical Informatics (AREA)
• Veterinary Medicine (AREA)
• Physics & Mathematics (AREA)
• General Health & Medical Sciences (AREA)
• Biophysics (AREA)
• Pathology (AREA)
• Biomedical Technology (AREA)
• Heart & Thoracic Surgery (AREA)
• Animal Behavior & Ethology (AREA)
• Molecular Biology (AREA)
• Surgery (AREA)
• Signal Processing (AREA)
• Physiology (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Artificial Intelligence (AREA)
• Psychiatry (AREA)
• Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
• Magnetic Resonance Imaging Apparatus (AREA)
• Apparatus For Radiation Diagnosis (AREA)

Priority Applications (6)

Applications Claiming Priority (2)

Publications (1)

Family

ID=50159299

Family Applications (1)

Country Status (6)

Families Citing this family (3)

Citations (5)

Family Cites Families (9)

• 2013
  • 2013-12-13 JP JP2015550172A patent/JP6235608B2/ja active Active
  • 2013-12-13 US US14/654,570 patent/US9826937B2/en active Active
  • 2013-12-13 WO PCT/IB2013/060923 patent/WO2014102653A1/en not_active Ceased
  • 2013-12-13 EP EP13831907.4A patent/EP2938247B1/en active Active
  • 2013-12-13 RU RU2015131831A patent/RU2685352C2/ru not_active Application Discontinuation
  • 2013-12-13 BR BR112015015522A patent/BR112015015522A2/pt not_active Application Discontinuation

Patent Citations (5)

Also Published As

Similar Documents

Legal Events