Development of a Photoacoustic Sensor Prototype for Non-Invasive In Vivo Glucose Concentration Measurement

Abstract

Non-invasive monitoring of biomarkers such as glucose is increasingly in demand due to the growing interest in preventive and personalized medicine. However, this field faces significant scientific challenges. Among the existing approaches, optical methods for in vivo measurements stand out. The main obstacle is signal distortion caused by light scattering and absorption in superficial skin layers and deeper structures. A promising way to overcome this limitation is the photoacoustic method, which enables depth-resolved tissue probing and signal acquisition from specific regions, such as blood vessels. This work presents a prototype of an optoacoustic device developed by the authors, which successfully registered signals in vivo. This development serves as a starting point for creating a device capable of non-invasive blood analyte measurement.

At the first stage of photoacoustic sensor development, a series of experiments were conducted using tissue-equivalent phantoms to determine the optimal characteristics of laser sources and an ultrasound receiver. A broadband ultrasonic detector was used to identify the central frequency providing maximum acoustic response amplitude. The experiments were performed at the laboratory of medical equipment for in vitro diagnostics at the Moscow Institute of Physics and Technology. A nanosecond laser with a pulse duration of 5 ns was employed. The operating wavelength (1550 nm) was selected based on the maximum glucose absorption. Under the laser pulse, analyte molecules absorbed light energy, part of which converted into heat, causing rapid heating and thermal expansion, leading to ultrasound wave generation. Signal registration was performed at five different phantom points (corners and center) with 10 measurements each, totaling 50 measurements per concentration level. A clear linear relationship was established between the acoustic signal amplitude and glucose concentration in the range of 50–200 mmol/L.

For in vivo measurements, a laboratory prototype was assembled combining two laser sources (1064 nm and 1550 nm, pulse duration up to 20 ns, pulse energy up to 1 mJ) and a 5 MHz piezoelectric ultrasound detector. The laser beams were focused into a single point in a special holder where the fingertip was placed. The 1550 nm beam spot diameter was 0.5 mm, while the 1064 nm beam spot was 2 mm. The ultrasonic receiver was positioned so its central axis crossed the beam intersection point. The measurement procedure involved placing the finger in the holder, filling the cavity with distilled water to ensure acoustic coupling, and sequentially recording photoacoustic signals from both lasers, followed by reference glucometer measurements within five minutes.

Over 200 paired measurements were collected from 12 volunteers. A glucose concentration calculation algorithm was developed based on two-stage signal processing. First, the 1064 nm signal was used to determine the target blood vessel depth and estimate blood volume. Then, the 1550 nm signal was recorded, and the response from the same depth was extracted and normalized to blood volume. This normalized value was compared with a calibration curve built from reference measurements. While some measurements showed high accuracy with deviations within 2–9%, the overall accuracy does not yet meet clinical requirements (ISO 15197), with outliers up to 44.4%. Nevertheless, the obtained results confirm the fundamental feasibility of the proposed approach. Future work will focus on prototype optimization, improved acoustic coupling, collection of additional experimental data, and refinement of calculation algorithms using reference instruments.

Speaker

Vladimir Kozlov
MIPT
Russia

Discussion

Ask question