Detección De Microplásticos En El Medio Marino Mediante Backscattering Ultrasónico De Alta Frecuencia

Marine microplastic pollution requires rapid, non-destructive, and in situ monitoring methods. High-frequency ultrasonic (HF-US) backscattering enables the detection of suspended particles and the estimation of their concentration without relying on optical systems or complex sample preparation. This Master’s Thesis explores the feasibility of such an approach, from instrumental design to quantitative calibration, within the framework of the OneBlue project. A compact platform was developed (Difrascope UT + HF transducer + peristaltic pump + MATLAB/Arduino control) together with an M-Scan processing pipeline consisting of the suppression of “vertical band” artifacts, enhancement by cross-correlation between columns with dB smoothing, and peak detection with event counting, producing a frame- by-frame series of particle counts. Experiments were conducted with polystyrene microspheres of 100–300 µm at three concentrations (11, 21, and 35 particles/µL), under three transducer angles (30°, 35°, 40°) and three flow velocities (10, 15, 25). Quantitative evaluation was based on a robust sliding-window counting indicator, combined with block-bootstrap resampling (B=2000) to estimate 95% confidence intervals and variances, and weighted linear regression (WLS) to derive the concentration–count calibration curve and its estimation formula with uncertainty propagation using the Delta method. The results showed that the optimal angle was 40°, as it provided a larger number of detections and a more favorable energy distribution compared to 30° and 35°. The flow velocity of 10v was discarded due to zero inflation and insufficient precision, while 25v demonstrated better sensitivity and dynamic range than 15v and was therefore adopted as the main experimental condition. A clear linear relationship between concentration and count was obtained at 25v with 𝑅𝑅w² = 0.958 and slope 𝛼𝛼 ≈ 1.724, along with an explicit calibration formula and its confidence intervals, with count uncertainties estimated through block bootstrap. These findings validate the feasibility of HF-US for detecting and quantifying microplastics (100–300 µm) under controlled conditions, as the hardware–software system produced stable counts, an interpretable linear calibration, and a reproducible statistical procedure. The study also proposes improvements, including rigid fixation at 40°, coupling and flow control, adaptive thresholds, statistical models accounting for overdispersion, and tests with additional concentration levels, in order to strengthen applicability across different media and advance toward online monitoring.