Summer Bursary Report: Victoria Walls
Over the coming weeks we'll be showcasing some of the experiences of students who have taken advantage of Christ Church's summer bursary programme. Summer bursaries give students funding to take part in educational and career development opportunities over the long vacation. Applications have just opened for 2022. To see the eligibility criteria and to apply visit the Summer Bursary page.
During my 12 week internship with the NIL group at the University of Oxford Engineering Department, I investigated the use of pulse oximetry within a smart mouthguard for use in contact sport. This was part of a larger project to obtain body measurements during sport. The aim of my internship was to obtain the heartrate of the user from the pulse plethysmography (PPG) sensor used in the device. Then to verify the estimate by comparing it against one from a gold standard pulse oximeter.
In order to do this, I had to decide which device to use as the gold standard. Some of the requirements were: it had to record the raw PPG signal, heart rate and pulse oxygen levels; it must be medical grade; must be able to be used when the subject is moving; and the cost must be taken into consideration. Considering these factors, the Creative PC-68B Bluetooth Enabled Wrist Pulse Oximeter appeared most suitable for our use.
We wanted to estimate the heart rate of the user using the PPG sensor, and so an algorithm had to be found. Researching potential algorithms revealed that a Fast Fourier Transform based algorithm could be best for our application and timescale. However, other, more rigorous methods, were also found and could have been implemented given more time. I then wrote the code to estimate the heart rate from our sensor based on the algorithm found in my research.
Since the PPG sensor was to be placed in the mouthguard, it was necessary to identify a suitable place to obtain the signal in the mouth. In order to do this, I researched potential sites documented in existing literature. From this, multiple potential sites were identified. Behind the teeth in order to take readings from the sub-branches of the Greater Palatine Artery was an ideal place, as having the sensor on the inside of the mouth could reduce the effect of environmental light. Another potential option was towards the front of the mouth and in front of the teeth. Whilst it is possible to obtain PPG signal from the teeth themselves (that the heart rate could be estimated from), the wavelengths of the LEDs on the sensor we used were not in the range to do so. In order to decide which site to use, comparison of signals obtained at each site were compared using our sensor.
The sensor (and other electronics) are embedded within the mouthguard, which is made of EVA plastics. Currently the EVA between the electronics and the mouth/gum is 1 mm thick. As the PPG sensor operates by measuring the reflected light from pulsing LEDS, introducing EVA between the sensor and the site of interest would likely have an impact on the signal. This was investigated by recording signals taken with no EVA, 1, 2, and 4 mm thick EVA. For the InfraRed LED signal the 1 mm EVA signal is more than 75% of the intensity of the signal with no EVA. For the Red LED there is more discrete variation between the 
intensities of the signals and when 2 mm is used, the signal intensity is higher than for the InfraRed. The error between the estimated heart rate and the gold standard was greater as the EVA thickness increased. The increase in error is small going from no EVA to 1 mm, but then drastically increases as the EVA increases to 2 mm and above. Further investigations will have to be carried out within the mouth, taking into account the site of measurement and the EVA thickness of the formed mouthguard. There may also be a change of the optical properties of the EVA after it has been heated and shaped into the mouthguard.
