
A NEW APPROACH TO EXTRACT FETAL ELECTROCARDIOGRAM USING AFFINE
COMBINATION OF ADAPTIVE FILTERS
Yu Xuan?1Xiangyu Zhang?2Shuyue Stella Li†2
Zihan Shen†3Xin Xie†1Leibny Paola Garcia2Roberto Togneri4
1University of California San Diego, 2Johns Hopkins University
3University of Chinese Academy of Sciences 4University of Western Australia
ABSTRACT
The detection of abnormal fetal heartbeats during pregnancy
is important for monitoring the health conditions of the fe-
tus. While adult ECG has made several advances in modern
medicine, noninvasive fetal electrocardiography (FECG) re-
mains a great challenge. In this paper, we introduce a new
method based on affine combinations of adaptive filters to
extract FECG signals. The affine combination of multi-
ple filters is able to precisely fit the reference signal, and
thus obtain more accurate FECGs. We proposed a method
to combine the Least Mean Square (LMS) and Recursive
Least Squares (RLS) filters. Our approach found that the
Combined Recursive Least Squares (CRLS) filter achieves
the best performance among all proposed combinations. In
addition, we found that CRLS is more advantageous in ex-
tracting FECG from abdominal electrocardiograms (AECG)
with a small signal-to-noise ratio (SNR). Compared with the
state-of-the-art MSF-ANC method, CRLS shows improved
performance. The sensitivity, accuracy and F1 score are
improved by 3.58%, 2.39% and 1.36%, respectively.
Index Terms—combination of adaptive filter, noninva-
sive FECG extraction, affine combination
1. INTRODUCTION
Fetal electrocardiography (FECG) is an important tool used
by clinicians to track fetal cardiac status [2] and can be ef-
fectively used to help clinicians make correct and responsive
decisions during pregnancy or at delivery. Currently, there
are two methods (Fig 1) to capture FECG: one is the inva-
sive scalp electrode [3] method, which directly measures the
FECG signal, but can only detect the FECG signal during de-
livery. Besides, the scalp electrode is invasive and may cause
serious harm to both the mother and the fetus. The other
method uses a non-invasive abdominal electrode [4]. An
electrode is placed on the mother’s abdomen to collect the ab-
dominal body surface signal. However, the abdominal body
?†Equal contribution in alphabetical order
Fig. 1: FECG monitoring a) Non-invasive b) Invasive [1]
surface signals of pregnant women are rather complex, in-
cluding not only the weak FECG and maternal electrocardio-
graphy (MECG), but also the mother’s respiratory noise and
industrial frequency interference. In addition, the amplitude
of MECG in the abdominal signal is often 2-10 times higher
than that of FECG [5], making the extracted FECG extremely
noisy.
In this paper, we introduce an affine combination ap-
proach for ECG separation using multiple adaptive filters.
Unlike previous combinations of multiple adaptive filters, the
FECG signal extracted by the affine combination is closer to
the true value and more complete. In addition, previous work
mainly focuses on exploring the combination of only LMS
filters with different step-size strategies. This is because these
filters allow the combination to minimize their own quadratic
errors by simply updating filter weights according to LMS
rules. However, it is known that the performance of single
Recursive Least Squares (RLS) is better than single LMS (and
its family) [6] on most denoising tasks because RLS is more
adaptable to non-stationary signals. The signal is generally
non-stationary in most practical application scenarios, and
the objective function of RLS is the exponentially weighted
arXiv:2210.11658v4 [eess.SP] 27 Feb 2023