Unlike magnetocadiography (MCG), ^{[1– 3]} noninvasive fetal electrocardiogram (ECG) detection promises to be a much cheaper way to monitor fetal status.

Fetal cardiac electrical activities pass through various body layers, and are finally picked up by electrodes on the pregnant abdominal skin. The abdominal ECG (aECG) record of a pregnant woman is composed of the maternal ECG component, the fetal ECG component, and various interferences (due to muscular and respiratory activities, etc.) and noises.^{[4– 7]} Among various recorded abdominal ECGs, those with greater fECG components will be more likely and easier to extract clear fECGs.^{[8, – 12]} The recorded abdominal ECGs seem quite different as the electrode location varies. It is therefore necessary to optimize the electrode locations to record the abdominal ECG signals with higher fECG signal-to-noise ratios (SNRs).

A number of electrode position schemes have been proposed for fetal ECG collection. In general, they can be divided into two groups. One is to place the electrodes around the maternal abdomen, ^{[13, 14]} and another is to place the electrodes on the abdominal surface.^{[15– 17]} Considering the latter group, the optimized electrode position scheme has not been obtained or confirmed based on theoretical analyses or experimental studies. Therefore, in this paper, some regularity on how to place electrodes on the maternal abdominal surface for higher fECG SNRs is investigated, based on a large amount of experimental data.

The rest part of this paper is arranged as follows. Section 2 describes how to collect the experimental data. Section 3 provides the statistics of fECG SNRs based on these experimental data. Section 4 gives a further discussion on the regularity of fECG SNRs. Section 5 ends with the conclusion.

Our data acquisition system is composed of a 1350P-ECG (a 12-lead electrocardiograph made by Nihon Kohden Corporation) and a personal computer (PC). The electrocardiograph acts as a data acquisition module and is connected to the PC with a dedicated USB cable. The sample frequency was set to 500  Hz, the cutoff frequency of the anti-aliasing filter was set to 75  Hz, and the switches of baseline drift suppression and EMG interference suppression were both set on.

For aECG collecting, three schemes of electrode placement were designed for ten electrodes, as illustrated in Fig.  1. In these schemes, six electrodes, which are originally thoracic electrodes, labeled from C1 to C6, were placed on six vertices of the hexagon, inscribed in a circumference with the pregnant woman's navel as its center and 10  cm as its radius, keeping fixed at these locations for each pregnant woman. Among the remaining four electrodes, which were originally limb electrodes, the electrodes labeled as F and R were fixed at the uterus fundus and pubic, respectively, for three schemes, while the other two labeled as L and N(RF) have different positions, as shown in Figs.  1(a)– 1(c).

Fig.  1.

	Figure Option ViewDownloadNew Window
	Fig.  1. Three placement schemes for the ten electrodes.

The experimental steps were as follows: (i) let the woman lie down on a bed and clean her abdominal skin with alcohol and apply saline water at the positions where the electrodes will be pasted (the purpose of this step is to remove the grease on the skin and enhance the conductivity of the electrodes); (ii) place the ten electrodes according to Figs.  1(a)– 1(c) respectively, and for each placement scheme, record 1– 2 groups of data with a duration of 24 seconds.

Experiments were carried out on 78 pregnant women in the Obstetrics Branch of Nanjing General Hospital of Nanjing Military Command. All together, 343 groups of aECG records have been collected. Among them, 300 groups were acquired from those pregnant women with fetuses in cephalic head position and 43 groups from those with fetuses in breech position. In this paper, these 343 groups of data composed the original aECG database.

In the above section, why do channels 1, 5, 9, 10, and 6 have much higher fECG SNRs than the rest? And, especially, why does the channel 1 has the highest fECG SNR of all? Perhaps we can explain these problems as follows.

First, the directions of these five channels are basically consistent with the fetal cardiac axis. As we know, the cardiac electric activity can be simplified as a dipole, with a variation in both the orientation and intensity of polarization. Signals picked up from the skin are in fact projections of the polarization vectors on the directions of channels. Since the polarization is time variable, the average orientation of the polarization vectors, known as the cardiac axis, is often considered. Generally, a higher channel amplitude can be achieved when the channel direction is more consistent with the cardiac axis. The typical normal direction of the cardiac axis is considered as moving from the right shoulder to the left leg.

In all our 274 groups of successful fECG exaction data, we know that, 91.6% result from the fetus in a cephalic position and the remaining are in a breech position. Therefore, we take the fetus cephalic position as an example. Figure  6 illustrates the relationship between fetal cardiac axis and channels 1, 5, 9, 10, and 6 under two different cases. In Fig.  6, the solid arrows represent the direction of channel directions with higher fECG SNRs and the dashed arrow represents the direction of the fetal cardiac electrical axis. We can see that in any cases of the fetal cephalic position, the channels with higher fECG SNRs have a feature in common that they are more consistent with the fetal cardiac electrical axis than the rest.

Second, the fact that the channel 1 has the highest fECG SNR of all, we think, comes from the farthest distance for two points F and R of channel 1. The distance between two points F and R is at least 20  cm longer than those of other channels. Although the abdominal surface potentials do come from the same fetal cardiac activity, they will have different phases for different points; i.e., different time delays. The greater the distance of the two points, the more time delays of two point potentials, which will lead to a bigger potential difference amplitude.

Fig.  6.

	Figure Option ViewDownloadNew Window
	Fig.  6. Direction relationship between five channels with higher fetal SNR and the fetal cardiac electrical axis in the cases of fetus (a) facing forward and (b) facing backward. The dashed line arrow represents the direction of the fetal cardiac electrical axis and the solid line arrows represent the directions of the five channels.

In this paper, using a standard 12-lead electrocardiograph, we collected the abdominal ECG signals from 78 pregnant women and then studied the regularity of the fECG SNRs related to electrode locations on the maternal abdominal surface. Consequently, five channels of signals are discovered to have higher fECG SNRs. These five channels also have a feature in common in that the channel directions are more consistent with the fetal cardiac electrical axis than the other channels. On the other hand, the bigger distance between two detecting points may be helpful in acquiring fECG signals with higher fECG SNRs. These findings must provide an important guidance for the placement of the electrode positions when detecting the fetal ECG.

The unique feature of our paper is that the original experimental data were expanded and integrated by the clever use of the difference method. In addition, the statistical result is also reasonably explained by the theories of the fetal cardiac electrical axis and signal phase delay.

A limitation of our work is that we still lack data from the fetus in breech position. To verify its clinical feasibility, more experiments must be conducted to expand the database.