On the performance of AS-LMS based Adaptive Filter
for Reduction of Motion Artifacts from PPG Signals
M. Raghu Ram
1
, K. Venu Madhav
2
, E. Hari Krishna
3
, Nagarjuna Reddy Komalla
4
,
K. Ashoka Reddy
5
1,2,5
Dept. of E&I Engg., Kakatiya Institute of Technology & Science, Warangal, Andhra Pradesh, India.
3
Dept. of ECE, KU CE&T, Kakatiya University, Warangal, Andhra Pradesh, India.
4
Dept. of Anesthesiology, Govt. MGM hospitals, Warangal, Andhra Pradesh, India.
Email:
1
ramcapri@yahoo.co.uk,
2
kotturvenu@yahoo.com,
3
hari_etta@yahoo.co.in,
4
sharanyanh@gmail.com,
5
reddy.ashok@yahoo.com
Abstract –
Photoplethysmographic (PPG) signal is invariably
corrupted with motion artifacts (MA) due to voluntary or
involuntary movements of the patient. PPG is a non-invasive
signal, used for the estimation of arterial blood oxygen saturation
(SpO
2
), which helps the physician to know
the hypoxic status of
patient during clinical investigations.
This paper presents an
efficient Adaptive Step- size Least Mean Squares (AS-LMS)
based adaptive filter for reducing the MA from corrupted PPG
signals.
The novelty of the proposed algorithm lies in the fact that
a synthetic noise reference signal for adaptive filtering,
representing MA noise, is generated internally from the MA
corrupted PPG signal itself instead of using any additional
hardware such as accelerometer or source-detector pair for
acquiring noise reference signal. Convergence analysis, SNR
calculations and Statistical analysis revealed that the proposed
AS-LMS algorithm has a clear edge over the Time-Varying Step-
size LMS (TVS-LMS) and Constant Step-size LMS (CS-LMS)
based adaptive algorithms for MA reduction from PPG signals.
Experimental results, for the PPG data recorded with different
motion artifacts (Horizontal, Vertical and Bending motion of
finger), demonstrated the efficacy of the proposed algorithm in
MA reduction and thus making it best suitable for real-time pulse
oximetry applications.
Keywords- PPG, MA, AS-LMS Adaptive Filter.
I. INTRODUCTION
Photoplethysmography, a noninvasive technique, is
frequently used in clinical investigations for the measurement
of heart rate and arterial oxygen saturation (SpO
2
). PPG is a
pulsatile waveform acquired from pulse oximeter using simple
inexpensive optical sensors (red and IR PPG sensors). The
simplest PPG sensor consists of an infrared LED and a photo
detector placed in a small plastic housing. The sensor is
applied to the skin by means of a double-faced adhesive ring.
The sensor can be either of transmitting type or reflecting
type. The pulsatile portions of red and IR PPG signals may be
used for reliable estimation of accurate SpO
2
. The part of
detected PPG signal due to the arterial blood appears pulsatile
in nature at the heart rate. The pulsatile portion of the PPG
arises due to the light passing through arterial blood and hence
has information contained in the arterial blood flow like; heart
rate, heart rate variability, respiration and blood pressure to
name a few [1]. Reliable and accurate estimation of SpO
2
requires clean artifact-free PPG signals with clearly separable
DC and AC parts. The main cause for deterioration of
accuracy in pulse oximeters is the corrupted portions created
in the detected PPG signals with voluntary or involuntary
movements (motion artifacts) of patient. While in process,
even slightest movement of patient disturbs the PPG signal,
resulting in inaccurate estimation of SpO
2
. Hence, reduction of
motion artifact (MA) has been a challenging pulse oximetry
problem, ever since its invention, for a reliable estimation of
SpO
2
.
Signal processing techniques are efficiently used for MA
reduction from the corrupted PPG signals. Since the frequency
spectrum of this noise (MA) overlaps frequency spectrum of
the desired signal (artifact-free PPG), traditional filtering
methods fail to remove the noise. To some extent, effect of
motion artifacts can be reduced by displaying the average
value of several SpO
2
readings, which is a typical procedure
followed by the manufacturers of commercial pulse oximeters.
To deal with this challenging problem, several methods
were proposed for MA reduction. The most common
technique employed for the MA reduction is the moving
average method [2]. But, it works well only for a limited range
of artifacts. Practically, the in-band noise due to MA
corruption can be successfully reduced with the use of
adaptive filters [3]-[5], by keeping in view of simplicity and
advantages like faster response time, avoidance of pulse
segmentation and ability to continue processing under MA
conditions. Use of a synthetic reference signal estimated from
the artifact-free part of the PPG signal [6] was also reported
for reducing motion artifact. Significant improvements were
observed by applying time-frequency methods like smoothed
pseudo Wigner-Ville distribution [7], making use of the non-
stationary nature of the PPG signals. In this paper, we present
a simple and efficient adaptive technique, focusing on accurate
reconstruction of the signal waveform, which automatically
help in better estimation of SpO
2
, rather on attempting to
stabilize saturation estimates. The proposed adaptive
technique utilizes AS-LMS algorithm for MA reduction. In
general, extra hardware like additional source-detector pair or
accelerometer are employed for generating a reference signal
for adaptive filtering application to PPG signals [3]-[5]. The
novelty of the proposed method lies in the very fact that a
synthetic noise reference signal is generated from the recorded
PPG data itself.
II.
MOTION ARTIFACT REDUCTION
Adaptive filters are proved to be best filters in noise
cancellation by self adjusting the filter coefficients based on
an adaptive algorithm, or close-loop adaptation. Different
algorithms [8] may be employed to adapt the weights w of the
filter, with an attempt to minimize the Mean Square Error
978-1-4244-7935-1/11/$26.00 ©2011 IEEE