Ultrasonic and Artificial Intelligence-Based Detection of Microplastics in Aquatic Environments

The water pollution by microplastic places a great danger to the wellbeing and ecological setup, requiring a more enhanced detection modality that extends beyond conventional methods in terms of scalability and performance. The conventional techniques of analysis, such as Fourier-transform infrared (FTIR) and Raman spectroscopy can provide highly accurate data but are limited by complexity in operation, high costs, and throughput. To address these deficiencies, this paper examines a synergistic model connecting ultrasonic sensing with artificial intelligence (AI) which includes machine learning and deep learning. Performing ultrasonic manipulation makes it possible to characterize microplastics in real time and in a non-invasive manner even in turbid conditions, whereas AI-based algorithms make it possible to identify, classify, and fuse multi-modal data to perform polymer identification and classification. As demonstrated by experimental findings and literature review, image-based AI models (CNNs, YOLOv5, nano-DIHM), as well as AI-enhanced spectroscopy, are more accurate and reproduceable in the laboratory and in the field. The main obstacles have not been eliminated, the most obvious ones being the lack of data and two-sided nature of the environment and engineering limits on large-scale implementation. Nevertheless, the offered hybrid solution shows great perspectives of scalable in-situ monitoring of microplastics, which has an implication on environmental stewardship and regulatory applications. The research should be improved with developing interoperable datasets and strong, interpretable models in the future to promote real-time detection of microplastic in the water.