PhD opportunity

Trustworthy AI-Enabled Multispectral Photoacoustic Imaging for Real-Time Therapeutic Monitoring

Funding availability

Unfunded

Application deadline

31 January 2027

  • Funding – China Scholarship Council PhD Programme
  • Application deadline 31st January 2027
  • Standard Entry date - September 2027
  • Applicants are expected to have a degree (equivalent of Honours or Masters) in a relevant discipline.

Innovative ultrasound-based therapies (e.g. focused ultrasound, sonodynamic therapy) enable non-invasive targeting of disease, but standard B-mode ultrasound lacks direct measures of tissue oxygenation and perfusion. Multispectral photoacoustic imaging (MSPAI) can fill this gap. MSPAI illuminates tissue at multiple wavelengths and reconstructs optical-absorption maps, allowing direct visualization of hemoglobin concentration and oxygen saturation (sO₂) in real time【2†L53-L61】【21†L141-L149】. These are critical functional biomarkers for monitoring tissue response during therapy. 

The aim is to develop a trustworthy AI-enabled MSPAI pipeline for real-time therapeutic monitoring. Quantitative biomarker estimation is challenging due to unknown light fluence, tissue motion, noise and system variability. We will combine physics-based modelling and machine learning: apply fast spectral unmixing and fluence correction for multi-wavelength data, and use co-registered ultrasound for motion compensation. An ML-based quality and uncertainty estimator will identify and mask unreliable pixels (e.g. low-SNR regions)【10†L281-L290】. For example, Qin et al. demonstrated that residual-based noise masking improves sensitivity to oxygenation changes by excluding noisy data【10†L281-L290】. We will also explore uncertainty-aware approaches (e.g. invertible neural networks) to output full posterior distributions of biomarkers (sO₂, Hb/HbO₂) rather than just point estimates【19†L19-L28】. 

For validation, we will image multi-wavelength tissue phantoms and ex vivo samples during controlled ultrasound exposures. Phantoms will have known optical properties (e.g. the OpenQPAIData dataset of 30 annotated phantoms【2†L68-L74】) to provide ground truth for calibration. In experiments, focused ultrasound or sonodynamic pulses will induce changes in vascular oxygenation; the MSPAI system must detect and quantify these dynamic changes. We will benchmark quantitative error, repeatability, latency and generalisation across conditions and labs, comparing MSPAI-derived biomarkers to independent reference measures (e.g. oximetry or histology). 

The expected outcome is a proof-of-concept real-time monitoring platform delivering calibrated MSPAI biomarkers with uncertainty estimates during therapy. Such a platform could enable adaptive ultrasound treatments guided by direct physiological feedback, improving safety and efficacy. The student will gain interdisciplinary training in AI,

How to apply

Please apply through the link below, you may contact the principal project supervisor to discuss your interest further, see supervisor details below.

Apply now

Supervisors

Principal supervisor

Related PhD programme