Machine learning based dual-modality UT-ECT tomography system with acoustic beamforming for 3D lab-on-chip tomography imaging

This PhD project explores the development of a dual-modality Electrical Capacitance Tomography (ECT) and Ultrasonic Tomography (UT) system, enabled by piezoelectric micromachined ultrasonic transducers (PMUT), to achieve enhanced object localisation and image reconstruction through acoustic beamforming. PMUTs are utilised not only as ultrasonic transmitters and receivers but also as capacitive electrodes, allowing simultaneous acquisition of ECT and UT data from a sensor array.

In conventional ECT or UT systems, high-strength signal acquisition requires parallel alignment between transmitter and receiver electrodes, which is a constraint that limits performance in array-based configurations. This project addresses this challenge by employing acoustic beamforming to steer ultrasonic waves toward targeted locations in the sensing region, enabling strong signal acquisition even between non-parallel electrode pairs.

This technique enhances the system’s ability to detect and localise floating or moving objects in a lab-on-chip (LoC) platform, hence producing accurate 3D image reconstruction. Beamforming is achieved by phase-controlled activation of multiple PMUT transmitters, with the direction optimised using preliminary object location estimated from capacitance measurements.

The objectives of this project are to (1) simulate and design beamforming protocols for UT and generate sensitivity maps, (2) apply machine learning to adaptively select and phase-control PMUT electrodes based on capacitance data, (3) design and fabricate a data acquisition system with beamforming capability, and (4) demonstrate beamforming-enabled lab-on-chip 3D tomography imaging.

This research aims to advance compact, intelligent imaging systems for applications such as multiphase flow monitoring, surface and subsurface defect detection and biomedical diagnostics.