Dataset for high-frequency ultrasound–based microplastic identification and size estimation

This dataset contains high-frequency ultrasound measurements acquired from controlled experiments on microplastic microspheres of different materials and size ranges. The data support research on ultrasound-based detection, material identification, and size estimation of microplastic particles. Raw data The raw data consist of three-dimensional tensors representing the spatial and temporal structure of the recorded ultrasound signals, with dimensions corresponding to the lateral scan coordinates (x, y) and time (t). These tensors were acquired over a defined scan area for samples containing microplastic microspheres. The raw radio-frequency (RF) ultrasound signals are provided in MATLAB (.mat) format and are stored in the raw_data.zip archive. These files contain the original, unprocessed measurements recorded during the experiments. Processed and labeled data Particle-specific signals isolated using a peak-based extraction procedure are stored in the all_labeled_signals.csv file. This file contains signal representations derived from the raw measurements together with associated material and particle size labels. Each signal entry is assigned a unique particle identifier, which enables signals originating from the same particle to be grouped and traced back to the corresponding raw measurements. Intended use The dataset is intended to support the development, evaluation, and benchmarking of machine-learning methods for microplastic characterization using ultrasound.