Adaptive filtering system and method

Details Adaptive filtering system and method Adaptive filtering system and method

1999-09-23

2002-11-26

Isen, Forester W. (Department: 2747)

Electrical audio signal processing systems and devices

Acoustical noise or sound cancellation

Counterwave generation control path

C381S094700, C381S071110

Reexamination Certificate

active

06487295

ABSTRACT:

BACKGROUND
1. Field
The present invention relates to adaptive filtering and, more specifically, to an adaptive filtering system and method for reducing interference in a medical diagnostic signal.
2. Discussion of the Related Art
Medical monitoring equipment is often used in relatively “noisy” environments. Electrical noise can corrupt signals processed by the medical monitoring equipment. For example, signals from an electrocardiogram (ECG or EKG) are often used by medical professionals to diagnose heart disorders or ailments. The ECG signals often are corrupted by electrical noise from adjacent electrical supply and power lines in the vicinity of where the equipment is being used. In particular, because the desired electrical signals in an ECG are very low voltage, they are highly susceptible to noise introduced from different sources including electrical supply and power lines.
Because a normal ECG signal has frequency components which are the same frequency as common power lines, for example, interference from the power lines which occurs on the ECG signal can be difficult to separate from the desired ECG signal.
To reduce the noise in a desired ECG signal, narrow band notch filters are often used. However, the use of narrow band notch filters is of limited utility. For example, on one hand, such a notch filter must be very selective, having a very thin bandwidth in order to avoid filtering any of the desired components of the ECG signal. This requires a high order filter which requires many components or a relatively lengthy processing time. In addition, the frequency of a power line signal typically varies somewhere around 60 Hz (50 Hz in Europe) and accordingly, the noise component may vary. Accordingly, on the other hand, a narrow band filter having a fixed notch frequency has to be relatively broad banded to reduce the noise induced by the power line. In addition, although the power line frequency in the United States is relatively fixed and stable at 60 Hz, the power line frequency in European countries can vary substantially from country to country and even from day to day in some countries. Accordingly, fixed narrow band filtering techniques are of limited utility.
Another type of filtering technique is an adaptive filtering technique. An example of an adaptive filter is shown in
FIG. 1A. A
reference signal r is correlated to the noise and is filtered by a finite impulse response (FIR) filter
10
. Filter
10
has adaptive parameters which very the frequency of the signal being filtered based on feedback signal f. The filtered noise correlated reference signal r is then subtracted from the input signal s. Using this technique, the feedback adjusts the parameters of filter
10
so that undesired noise in the output signal is minimized. In this way, a filter with large stopband damping can be provided with minimal damping in the passband. Furthermore the frequency of the noise makes no difference, so that adaptive filtering finds some use in those countries where power line frequency varies considerably.
Various algorithms can be used to update or adjust the parameters of filter
10
. One simple method is referred to as minimizing the least mean square (LMS) of the output. To achieve this, filter parameters can be updated in the following way:
W
n+1
=W
n
+&mgr;y
(
n
)
r
*(
n
) (* designates complex conjugate)  (1)
where W
n
, is the parameter array at sample number n and &mgr; is the step size used to determine how fast the filter adapts.
In the adaptive filter shown in
FIG. 1A
, the purpose is to filter the noise correlated reference signal r so that it resembles the noise portion of the ECG+noise signal s. Then, by subtracting the filtered noise correlated reference signal r′ from the ECG+noise signal s, effective cancellation of the noise can be achieved. Although such an adaptive filtering technique may be effective for certain applications, such technique is not ideal and difficulties still remain.
An ECG output signal is shown in FIG.
1
B. An ECG waveform for a single beat is typically referred to as a PQRST complex. At initiation of a heart beat, a P wave appears. This corresponds to activity in the artia. The QRST complex then follows the P wave. The QRST complex corresponds to ventricular activity. More specifically, it is generally understood that the QRS portion of the wave corresponds to electrical activation of the ventricles and the T wave corresponds to their electrical recovery. The ST segment is relatively quiescent.
A typical ECG signal contains a 60 Hz component during a limited period of time. That is, a 60 Hz component is present in the QRS complex. During this short period the reference signal is highly correlated with the pure ECG signal. The adaptive filter coefficients are updated incorrectly. This results in a 60 HZ disturbance in the ST-segment before the coefficients readjust. Accordingly, if a filter is used to remove 60 Hz noise, a portion of the actual ECG signal itself is removed, thus an error occurs when using filter
10
, resulting in disturbance in the desired ECG signal. For example,
FIG. 2A
shows an ECG signal with simulated 50 Hz power line interference added.
FIG. 2B
shows the result of using filter
10
on the signal shown in FIG.
2
A.
FIG. 2C
shows a closer look at a ST-segment of the filtered signal, the ST-segment being corrupted by 50 Hz interference.
Accordingly, a need exists for a efficient system for eleminating noise in an ECG signal.
SUMMARY OF THE INVENTION
A signal filter includes a first adaptive filter receiving a desired signal containing a noise component and a first noise correlated reference signal, and outputting a first filtered output signal. A second adaptive filter, receives the desired signal containing the noise component and a second noise correlated reference signal and outputs a second filtered output signal. Signal filter parameters of the second adaptive filter are controlled based on the first filtered output signal and the second filtered output signal. The first and second noise correlated reference signals may be the same. The signal filter parameters of the first adaptive filter are adapted according to the formula W
1,n+1
=W
1,n
+&mgr;y
1
(n)r*(n), where n is the sample number and &mgr; is the step size. The signal filter parameters of the second adaptive filter are adapted according to the formula W
2,n+1
=W
2,n
+&eegr;(n)&mgr;y
2
(n)r*(n), where n is the sample number, &mgr; is the step size and &eegr; is a weighted signal based on the first filtered output signal. The first adaptive filter may include an adder, a finite impulse response (FIR) filter and an adaptive processor for controlling the adaptive parameters of the FIR filter. The second adaptive filter may include an adder, a second finite impulse response (FIR) filter and a second adaptive processor for controlling the adaptive parameters of the second FIR filter. A mathematical operation unit can be provided for outputting a weighted output signal based on the first filtered output signal, the weighted output signal being provided to the second adaptive filter as the first filtered output signal.
According to another aspect, a signal filtering method includes receiving a desired signal containing a noise component and receiving a noise correlated reference signal. Filtering of the noise correlated reference signal is performed using a first set of adaptive parameters to provide a first filtered noise correlated reference signal. The first filtered noise correlated reference signal is subtracted from the desired signal containing the noise component to provide a first output signal. The first set of adaptive parameters are controlled based on the first output signal. Filtering of the noise correlated reference signal is performed using a second set of adaptive parameters and a signal corresponding to the first output signal to provide a second filtered noise correlated reference signal and the second filtered noise correlated reference signal

Affiliated with

Also associated with

Rating

Adaptive filtering system and method does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Adaptive filtering system and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Adaptive filtering system and method will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2935141