Negar Tavossolian is Associate Professor of Electrical and Computer Engineering at Stevens Institute of Technology, Hoboken, NJ. She led a project to develop a lightweight, non-invasive sensor to monitor fetal heart rate and motion that could sense abnormalities warning of possible stillbirths.

Tech Briefs: What gave you the idea you could monitor fetal heart rate and movement using vibration sensors?

Professor Negar Tavassolian: For a few years, we have been using the very small motion sensors that are embedded in cell phones as vibration sensors to monitor adult heartbeats. We convert the motion data into information about cardiac activity by using signal processing algorithms. We then decided to attempt using that technology to monitor the fetal heart inside the abdomen of a pregnant woman.

Tech Briefs: It seems to me it would have to be much more sensitive for the fetal heart rate.

Prof. Tavassolian: Yes, it has to be a lot more sensitive. For adult monitoring, we use one sensor on the left side of the chest wall, on top of where the heart is located. However, for this experiment, we placed several sensors around the mother’s abdomen, and then with our signal processing method, combined the vibrations from the different locations to extract the heartbeat information.

Tech Briefs: Could you tell me something about how the sensor works.

Prof. Tavassolian: It’s a vibration sensor — on each tiny patch, there is one sensor to measure linear acceleration, and another to measure rotational speed. Together, they form a sensor that detects any kind of motion.

Tech Briefs: How do you distinguish between heart rate and fetal movement?

Prof. Tavassolian: Currently, we are trying to distinguish them through the signal characteristics. Heart rate is usually periodic and has a certain frequency, while fetal movement is relatively intermittent. There are also differences in the signal strength, frequency, and pattern between these two types of signals. Our algorithms extract periodic signals, so in that way we can obtain information related to heart rate. On the other hand, fetal movement might happen suddenly with a different pattern, and a different frequency and level. At the moment when you record the fetal heart rate, the fetal movement might not be strong but when fetal movement does occur, it will probably have a stronger signal, and a different type of amplitude shape.

Tech Briefs: At about what stage in pregnancy does this start being used?

Prof. Tavassolian: Basically, we want to monitor the heart after week 28, but for this pilot study, we targeted later stages — about 39 or 40 weeks. At that stage, the is quite strong.

Tech Briefs: Is there data about the stage at which stillbirths tend to happen?

Prof. Tavassolian: By definition, it happens when the fetus is relatively developed and is born dead. So, by definition, it happens in the later stages — there are some population statistics on that — It can generally occur at about 20 weeks or after. It can happen when the baby is about to be delivered or is still under development. Problems usually start because of malnutrition, depression, or other health conditions of the mother and the fetus. There are so many factors that we really cannot generalize to give a number to determine where or when this happens. Before 20 weeks, it’s defined as a miscarriage — after that it’s defined as stillbirth.

Tech Briefs: How is the data communicated?

Prof. Tavassolian: What we did was to log it in to a memory card. Currently there is no wireless communication between the prototype and any other device. However, in the future, we expect that this system will be able to communicate with a smart phone or some other handheld device, perhaps via Bluetooth, so that it can provide real-time data to the wearer and also transmit important data to the cloud for healthcare providers. Since these are really tiny sensors, very low-cost, lightweight, easy to apply, and don’t need electrodes, we can attach them to a strap that a woman could wrap around her abdomen. This would enable a pregnant mother to go outside and do daily tasks, while at the same time providing continuous monitoring.

We could do some initial processing at the user’s end to alert the mother if there are any abnormalities in the heartrate or fetal movements. We could also provide a complete record of data for the doctor to access and analyze.

Tech Briefs: Do the signal frequency and amplitude fall within fairly narrow limits from person to person.

Prof. Tavassolian: It’s relatively consistent. If we talk about the frequency component it might be similar, however the fetal heart rate is subject to variance from case to case. In this study, we focused on late pregnancies — at that time, the fetal heart-rate in beats per minute (BPM) has an expected range. The baby should not fall far above or below these accepted levels.

Tech Briefs: How does the accuracy of your device compare to fetal cardiotocograms (f-CTG)?

Prof. Tavassolian: The reliability of the signal — the statistical consistency of the accuracy — is within the same range as the f-CTG. The target of our future research is to improve our accuracy, so that when our results are compared to a gold reference like the (invasive) fetal scalp ECG, our performance will be shown to be comparable to the devices currently being used in the clinic.

Tech Briefs: Is change of data over time more important than absolute accuracy?

Prof. Tavassolian: Yes, monitoring the trend is actually more important than monitoring the absolute value.

Tech Briefs: Do you have a sense of when this might be put into actual practice?

Prof. Tavassolian: No, we haven’t really. One main thing we want to do is to combine these kinds of modalities with some other wearable sensing techniques. So, what we envision is not just to monitor the fetal heartbeat — we want to give a more complete overview of the health status of the fetus and the mother. We have a lot of experience using sensors for monitoring different vital signs in adults: heartrate, respiration rate, even blood pressure. Our goal is to monitor the maternal and fetal health with one product. We will use artificial learning to improve the accuracy, especially of the fetus heartrate evaluation, so I think it will be a few years before we have a final product on the market.

An edited version of this interview appeared in the October Issue of Tech Briefs.