A elaborate account of the proposed algorithm, which includes FECG signal extraction and FHR and MHR sensing, are presented in this chapter. In order to formalize the algorithm, the execution utilizing Matlab ( book ) and Simulink ( block ) are besides described.
Algorithm for off line processing
A new algorithm has been developed for this work in order to treat multiple-lead or two-lead with the possibility to treat single-lead abdominal signal. The FECG signal extraction consists of four phases. The First phase is preprocessing which involves remotion of the DC signal and base line wander utilizing zero mean and bandpass filter. The 2nd phase is the maternal extremum sensing, together with the data format of the MQRSW. The 3rd phase is to pull out the foetal signal utilizing ANC technique. In the 4th phase, the postprocessing technique is scaled down the MQRS composites in the extracted signal by using the window signal created in the 2nd phase.
The FHR and MHR sensing techniques, which are threshold-free, can be performed in two stairss. First measure is to plan the moving interval harmonizing to the lower limit and maximal RR intervals of the normal FHR and trying frequence. In the 2nd measure the first foetal extremum is detected, at the same clip the traveling interval is updated in order to observe the following foetal extremums. Finally, standards about amplitude and backward hunt are settled to rectify false sensings ( overlapped extremums ) utilizing window signals. The developed algorithm is illustrated in Figure 4.1 and discussed in item in the undermentioned subdivisions.
Figure 4. 1 Phases and stairss of the proposed algorithm
The preprocessing phase consists of the remotion of the DC signal, baseline wander and the power line intervention. The observation signals X1, X2 and X3 in the preprocessing phase are acquired from the maternal venters.
Where Xn is the observation signal. In order to take the DC of the signal, each observation signal is made zero average by using
( 4.1 )
Baseline wander is caused by the patient ‘s external respiration or motions during entering. The frequence of the baseline wander due to take a breathing is in the scope of 1 Hz and the EMG noise ( artefacts of muscular contractions ) is characterized by comparatively high frequence noise. Hence the recorded signals ( informations ) were passed through FIR overacting bandpass filter with the cutoffs at 40 Hz and 4 Hz which is the bandwidth of involvement for FECG monitoring. The frequence response of this set base on balls filter is shown in Figure 4.2.
Figure 4. 2 Frequency response of the bandpass filter
The rating of the preprocessing phase can be performed by comparing the input signal with the filtered signal. Figure 4.3 clearly present the ability of this phase to take or stamp down the unwanted frequences. The input signal ( AECG ) is shown at the ( a ) and the same signal after pre-processing is shown at the ( B ) of the Figure 4.3.
Figure 4. 3 ( a ) AECG signal and ( B ) preprocessed signal
Maternal Peak Detection
The MQRS extremums are used to make the MQRSW and cipher the MHR. For this intent a new amplitude threshold-free sensing algorithm is developed. In order to make the MQRSW the filtered Y2 ( illustrated in Figure 3.4 ) signal is used to observe its MQRS extremum. The MQRS extremums sensing algorithm depends on the maternal RR intervals of the normal MHR and trying frequence as shown in Figure 4.4.
Normal bosom round
Max. & A ; min. RR Time intervals
Traveling interval borders
First peak sensing
Updating traveling interval
Detecting the following extremum
Figure 4. 4 Block diagram of the MHR sensing algorithm
Figure 4.5 shows an illustration of the maternal RR traveling interval and its relation with the MHR which should be calculated earlier. In add-on to the normal MHR, the execution of this algorithm depends besides on the sampling frequence. The maximal normal MHR is about 100 beats per minute ( BPM ) and the lower limit is about 60 BPM. The sampling frequence is chosen to be 256 Hz in this work. The RR interval and its two-base hit can be calculated as follows:
( 4.2 )
Therefore, the maximal normal RR interval is about 250 samples ( its two-base hit is 500 samples ) and the lower limit is about 150 samples ( its two-base hit is 300 samples ) .
Figure 4. 5 The Maternal RR traveling interval
The RR intervals are ranged between 150 – 250 samples, and the scope of dual RR interval is 300 – 500 samples for a sampling frequence of 256 Hz. Hence in Figure. 4.5, the maximal RR interval is shown at the top, the minimal RR interval at the underside and between them an illustration of medium RR interval.
Let all maximal, medium and minimal RR intervals start at the same get downing extremum MP ( I ) , so the get downing point of the traveling interval ( SMI ) is chosen to be to account for the QRS composite after the detected extremum by 17 samples. The terminal point of the traveling interval ( EMI ) must be in a location that is greater than the location of the extremum ( A ) of the maximal RR interval and smaller than duplicate the minimal RR interval or before Peak ( B ) . Thus the EMI is chosen to be equal to. Within these bounds merely one extremum ( A, C, or D ) can be detected in every moving interval ( MI ) as shown in Figure. 4.5. Detecting the extremum in the MI is executed by taking the upper limit,
( 4.3 )
Once the extremum is detected, the MI will be shifted frontward after the detected extremum by 17 samples for SMI and by 290 samples for EMI in order to observe the following extremum between the borders of every new traveling interval. The rule of the peak sensing is illustrated in Figure 4.6. This rule does non necessitate any old cognition about the extremum threshold doing the sensing of low computational complexness and robust.
Figure 4. 6 Maternal extremum sensing utilizing MI.
The RR interval, i.e. the reciprocal of the bosom rate of the kth rhythm is defined as
( 4.4 )
and the MHR can be calculated as
( 4.5 )
where = 256 Hz is the sampling frequence.
At the same clip the same method can be applied to observe the ( negative ) minimal maternal signal points of index that are downward warps ( Q or S ) of the maternal QRS composite in every RR interval by
( 4.6 )
Creation of MQRS Window
The chief trouble in the fetal extracted signal is the residuary MQRS extremums particularly that are ( which are ) overlapped with the foetal extremums. Many techniques are proposed antecedently taking at optimum solution to divide the maternal residuary extremums. In this algorithm a new sweetening technique is developed, that is the MQRS window ( MQRSW ) . The design of this MQRSW depends on the maternal extremums detected in the old phase, where the indexes of these extremums are used as a centre of MQRSW. The MQRS composite is captured within a MQRSW, which is defined by taking 13 samples before, and after every detected maternal extremum found in the input signal Y2. Figure 4.7 explains how the constituents of MQRS complex inside the MQRSW are captured and remain constituents outside the MQRSW are eliminated ( zero padded ) .
The MQRS complex interval, which is represented by the MQRSW, is given by:
( 4.7 )
where s =13 is the figure of samples as shown in Figure 4.8 and the window signal can be defined as:
( 4.8 )
where is the peak index.
I = i+1
Y2 ( I ) = Mp ( I )
Y2 ( i-s ) = 0
I = 0
H = a?z
I & gt ; H
I & gt ; h + 2s
Figure 4. 7 Flow chart for the formation of the MQRSW
Figure 4. 8 An illustration of MQRSW
FECG Extraction Technique
Adaptive filters minimize the difference between the end product signal and the desired signal by changing their filter coefficients ( Jimenez et al. 2000 ) . One of the chief applications of adaptative filters is noise remotion utilizing adaptative noise canceling ( ANC ) .
In ANC, the end is to take the noise signal from the measured signal by utilizing a mention signal that is extremely correlated with the noise signal. Figure 4.9 shows the ANC system execution for the FECG extraction. A signal FECGp is transmitted over a channel to a detector that besides receives a noise MECGp, pi?Z [ 2,3 ] uncorrelated with the signal. The combined signal and noise form the primary input to the canceller. A 2nd detector receives a noise uncorrelated with the signal but correlated in some unknown manner with the noise. This detector provides the mention input to the canceller.
pi?Z [ 2,3 ]
Figure 4. 9 Adaptive noise canceller system
The and ( , ) are the signals acquired from the maternal venters. The noise is filtered to bring forth an end product that is every bit near a reproduction as possible of. This end product is subtracted from the primary input to bring forth the system end product.
( 4.9 )
where is the end product of the adaptative filter. Squaring both sides of Equation ( 4.9 )
( 4.10 )
Using outlooks on both sides of Equation 4.10
( 4.11 )
As is uncorrelated neither with nor with so the last term is zero. Finally,
( 4.12 )
The end of the adaptative filter is to minimise the average square mistake ( MSE ) of. This can be obtained iteratively to give the optimum solution when. If one knew the features of the channels over which the noise was transmitted to the primary and mention detectors, it would theoretically be possible to plan a fixed filter capable of altering into. The filter end product could so be subtracted from the primary input, and the system end product would be the signal alone.
However, the features of the transmittal waies are as a regulation unknown or known merely about and are seldom of a fixed nature, therefore the usage of a fixed filter is non executable. Furthermore, if a fixed filter were executable, its features would hold to be adjusted with a preciseness hard to achieve, and the slightest mistake could ensue in an addition in end product noise power. An adaptative filter differs from a fixed filter in that it automatically adjusts its ain impulse response. Adjustment is accomplished through an algorithm that responds to an mistake signal dependant, among other things, on the filter ‘s end product. Therefore with the proper algorithm, the filter can run under altering conditions and can readapt itself continuously to minimise the mistake signal.
After extraction utilizing ANC, it is noted that the maternal residuary extremums are still observed, which means that it is still hard to observe the foetal extremums. Hence an effectual sweetening technique is required to heighten the fetal extracted signals by rarefying other interfering constituents.
After ANC the foetal extracted signal corrupted with the maternal residuary extremum is fed to the postprocessing phase for sweetening. This phase consists of the undermentioned parts:
A. The MQRS scaling down window
The MQRSW is created as described in Section 4.2.3. Two scenarios are developed to heighten the foetal signal: foremost use the MQRSW as MQRS remotion window and 2nd use the MQRSW as MQRS scaling down window. The scaling down version is used in this work to avoid taking the maternal extremums, which are overlapped with foetal extremums, which need to be detected subsequently.
An illustration of the scaled down signal is shown in Figure 4.10 ( B ) . The window signal is used to rarefy the maternal residuary extremums in the fetal extracted signal. If peers to zero, so the fetal extracted signal remains the same. If the samples of are non zero, so the corresponding samples in the fetal extracted signal are scaled down by multiplying them by 0.1.
Figure 4. 10 ( a ) the window signal and ( B ) scaled down signal
B. IIR notch filtering
After scaling down the maternal residuary extremums in the extracted signal, a little sum of baseline wander has been observed. Therefore, a 2nd order IIR notch filter centered at 1 Hz is used to rarefy this baseline wander. The undermentioned transportation map is used for the filter:
( 4.11 )
To supply the needed noise fading, the value of is chosen I? to be 0.85. Figure 4.11 shows the frequence response of the IIR notch filter.
Figure 4. 11 Frequency response of IIR notch filter
C. Adjustment of the maternal QRS amplitude
The ensuing signal from the old phase is normalized and made absolute in order to use the foetal extremum sensing algorithm. The amplitudes of the maternal and the fetal overlapped extremums in the extracted signal demand to be adjusted, to maintain all amplitudes of the maternal extremums shorter than those of the foetal extremums. The first measure is to acquire the maximal value ( millivolt ) between next maternal extremums. Once the upper limit has been detected, the amplitude of the next maternal extremums in the extracted signal are adjusted to be 0.75 * millivolt. The maternal ( M ) and the possible overlapped ( F+M ) extremums ( labeled with black stars ) in the extracted foetal signal are adjusted as shown in figure 4.12 ( degree Celsius ) .
Figure 4. 12 ( a ) the window signal, ( B ) foetal signal and ( degree Celsius ) fetal
signal with adjusted maternal extremum amplitudes.
FHR Detection Algorithm
The rule of the foetal extremum sensing is based on the normal foetal bosom round, foetal RR interval and the sampling frequence similar to the maternal extremum sensing in subdivision 4.2.2. The FHR of involvement is between 93 to 180 beats per minute, Therefore the maximal normal RR interval is about 164 samples ( its two-base hit is 328 samples ) and the lower limit is about 85 samples ( its two-base hit is 170 samples ) .
Before explicating the sensing algorithm, the traveling interval should be defined as illustrated in figure 4.13.
The maximal RR interval is represented at the top of figure 4.13 and the minimal RR interval is represented at the underside. In between the other RR intervals within the frequences of involvement are shown. Assume the first foetal extremum FP ( I ) and its index ( I ) is used to specify the traveling interval ( MI ) . The SMI is chosen to be, that is after the QRS composite of FP ( I ) . The EMI must be in a location that is greater than the location of the extremum ( A ) of the maximal RR interval and smaller than duplicate the minimal RR interval or before Peak ( B ) , therefore the EMI is chosen to be.
Figure 4. 13 The moving interval for extremum sensing
Within these bounds merely one extremum ( A, C, D or E ) can be detected. Detecting the extremum in the MI is executed by taking the maximal value within the interval,
After the ( A, C, D or E ) is detected, by MI. The following extremum is detected by utilizing similar MI starting at five samples after the extremum. In this manner the foetal extremums are detected without utilizing amplitude threshold.
Peak Position Corrections
The extracted overlapped extremums are normally shifted around the original extremum because of the MQRS composite overlapped with the FQRS composite. In order to rectify the possible presence of overlapped extremums, a new rectification standard based on the window signal is developed. After peaks sensing, overlapped extremums which are shifted from their locations are corrected. All foetal extremums detected in Figure 4.14 ( B ) should hold value ( x=2 ) for the same index in Figure 4. 14 ( degree Celsius ) , if the value of the same index in Figure 4. 14 ( a ) is equal to nothing. The foetal extremums detected in Figure 4. 14 ( B ) should hold value ( y=1.45 ) for the same index in Figure 4. 14 ( degree Celsius ) , if the value of the same index in Figure 4. 14 ( a ) is non equal to nothing. The indexes of all value ( Y ) are corrected if the difference between RR intervals more than two samples before and after ( Y ) as shown in Figure 4. 14 ( degree Celsius ) .
Figure 4. 13 Peak place corrections
ALGORITHM IMPLEMENTATION USING SIMULINK
The proposed algorithm discussed in Section 4.2 has besides been implemented under MatlabA®/SimulinkA® for online FHR monitoring. Simulink theoretical account was created utilizing the blocks from Simulink Library and the blocks embedded with Matlab map. The full simulink theoretical account is shown in Appendix E. the chief parts are discussed in item as follows.
The preprocessing phase consists of the remotion of the DC signal, baseline wander and the power line intervention. Each observation signal is made zero average by using Equation 4.1. For this intent, an add/subtract block and a average block are copied into the theoretical account as shown in Figure 4.15.
The leftmost block in Figure 4.15 represents the three recorded abdominal signal which are saved in a computing machine file. The end product of the nothing mean map is connected to the MUX block to divide the ECG channels. Each ECG channel is so connected to the digital bandpass filter as described in Section 4.2.1. This phase is so ended by the buffering block of 1280 samples ( equal to 5 seconds ) as shown in Figure 4.15.
Figure 4. 14 Preprocessing phase
Maternal Peak Detection
The preprocessing phase is ended with buffer block and connected to Data Type Conversion blocks to work with informations in dual preciseness format. Y1, Y2 and Y3 are the preprocessed signals. Then two blocks are dragged from Simulink library: first block is for the maternal extremum sensing and 2nd block for creative activity of the MQRSW, which come in consecutive phases. In order to put to death the maternal extremum sensing phase, Embeded Matlab Function 1 is used as shown in Figure 4.16. The input to this block is the filtered signal from Y2 of the preprocessing phase. The Embedded Matlab Function 1 executes the sensing of the maternal extremums Mp, their indices im and the MHR as discussed in Section 4.2.2.
Figure 4. 15 Simulink Blocks for the window processing phase.
Creation of MQRS window
The window creative activity phase depends straight on the extremum indices. The Embedded Matlab Function 2 in Figure 4.16 is used to put to death the MQRSW creative activity undertaking as described in 4.2.3.
FECG Extraction Technique
The ANC technique discussed in subdivision 4.2.4 has been implemented here utilizing Simulink to pull out the FECG. In this phase the adaptative filter RLS blocks are connected to the preprocessing phase for FECG extraction as shown in Figure 4.16. The filtered signal from Y1 is fed as mention signal to the input port of the RLS Filter 1 and 2. At the same clip, Y2 is fed to the Desired ports of the RLS Filter 1 and to the Y2 port of the Embedded Matlab Function 1 in Figure 4.16. Y3 is fed as Desired signal to the desired port of the RLS Filter 2 in Figure 4.16
The end product of this phase is the foetal extracted signal, which comes out from the Error port. All the other variables from each block are maintained for processing in the undermentioned blocks.
This phase is applied to the fetal extracted signal by ANC to heighten the fetal extracted signal before foetal peak sensing phase. Postprocessing phase is shown in Figure 4.17 together with the foetal extremum sensing phases, which are all explained as follows.
A. Scaling Window Signal
All codifications in this phase are implemented utilizing. Embedded Matlab Function 3 as shown in Figure 4.17 to use the signal Xw to scale down the fetal extracted signal, from Error port of the RLS Filter 1. The rule of scaling down phase depends on the samples values of Xw ( if it is zero or non nothing ) as in Section 4.2.5 A
Figure 4.16 Simulink blocks for postprocessing and foetal extremum sensing phases
B. Notch filtering
After scaling down the maternal residuary extremums in the extracted signal, a little sum of baseline wander has been observed. Therefore, a 2nd order IIR notch filter centered at 1 Hz is used to rarefy this baseline wander. In Figure 4.17, Embedded Matlab Function 4 performs the features of the notch filter.
C. Maternal extremum amplitude accommodation
After the signal is filtered by 1 Hz IIR notch filter, the amplitude of the maternal extremums in the fetal extracted signal ( FECG2 ) would hold to be adjusted by Embedded Matlab Function 5 in Figure 4.17 as described in subdivision 4.2.5 C.
Amplitude Threshold Free Peak Detection
In Figure 4.17, Embedded Matlab Function 6 performs the foetal extremum sensing. The adjusted foetal signal from the end product port ( FECG3 ) of the Embedded Matlab Function 5 is fed to the input port ( FECG3 ) of the Embedded Matlab Function 6 in order to observe the foetal extremum. The rule of the foetal extremum sensing is explained in subdivision 4.2.6. After the peak sensing, the detected fetal overlapped extremums are corrected as explained in subdivision 4.2.7. Finally, the deliberate FHR can be monitored at the end product port ( FHR ) of the Embedded Matlab Function 6.
The development of the algorithm, which includes FECG signal extraction, FHR and MHR sensing, are presented. First, the algorithm was implemented utilizing Matlab 7.4 ( Section 4.2 ) . Second, the algorithm was implemented under Matlab simulink, in order to be used on line every bit good as in existent clip ( Section 4.3 ) . The complete execution of this algorithm under Matlab simulink, and the complete flow chart of this algorithm are given in Appendix C.