Measuring blood oxygenation in vivo by photoacoustic imaging

Lead Research Organisation: University College London
Department Name: Cell and Developmental Biology


Blood oxygenation is a critical measure of physiological function. However, there are very few techniques to reliably measure oxygenation non-invasively. In this interdisciplinary studentship, the student will undertake experimental imaging studies and computational modelling to develop a method to measure oxygenation using a novel technique called photoacoustic tomography. The technique will be tested in phantoms and the kidney, which has different regional areas of oxygenation related to its biological function and in models of hypoxia induced by chemical insults. If successful, this technique would revolutionise preclinical imaging techniques and could be applied to multiple physiological processes and pathological conditions.

PhD project: aims and description (limit 300 words)
Blood oxygenation is an important physiological indicator of tissue function and pathology, relevant to the study of pathophysiological processes such as angiogenesis in tumours, tissue inflammation and healing responses. Multiwavelength photoacoustic tomography (PAT) has the potential to provide high resolution 3D images of oxygenation at higher resolution than is currently possible. This would transform preclinical imaging, but one hurdle remains. There is a non-trivial spectroscopic step, and while promising mathematical solutions have been developed, translating these into robust procedures for preclinical application is challenging. To succeed, this project will require both the experimental and algorithmic aspects of the problem to be considered together.
Initially, optimal wavelength selection techniques will be explored numerically and experimentally in phantoms to determine their practical applicability to estimating oxygenation. For more accurate (but slower) estimation, nonlinear optimisations employing Monte Carlo optical models and k-space ultrasound models will be considered. Existing prototype algorithms will be developed and used with experimental data. BC will supervise this aspect in collaboration with experimentalists (led by PB).

After phantom experiments (see Rotation project), a series of in vivo preclinical experiments are envisaged. First, vessels in the mouse flank will be imaged, and oxygenation estimates compared with co-oximeter values. Following this validation, images of the normal mouse kidney will be obtained to examine the technique's physiological relevance. These aspects will be guided by DL. The kidney contains areas of hypoxia in the medulla which facilitate the reabsorption processes, therefore providing an ideal test-bed for PAT imaging. Chemical insults such as cobalt chloride will then be used in mice to induce hypoxia (which occurs in multiple pathological conditions) to examine whether PAT can detect oxygen changes in a pathological situation. The PAT estimates will be compared with current methods used to detect oxygen levels such as hypoxyprobe administration.


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
BB/M009513/1 01/10/2015 31/03/2024
1902537 Studentship BB/M009513/1 01/10/2017 30/12/2021 Ciaran Bench
Description A new technique for estimating vascular blood oxygen saturation from photoacoustic image data was developed and tested on realistic simulated images. Convolutional neural networks were trained to generate an image of the vascular sO2 in a tissue sample, from images of the tissue sample acquired at multiple wavelengths. The work was presented at a couple of conferences and was nominated for a best paper award at the largest photoacoustics conference.
Exploitation Route Our approach of using 3D convolutional neural networks to estimate vascular blood oxygen saturation from photoacoustic image data will likely be used by other groups seeking to acquire accurate estimates of blood oxygen saturation from photoacoustic images.
Sectors Pharmaceuticals and Medical Biotechnology