Tomographic imaging of flow and chromophore concentrations in biological tissue

Lead Research Organisation: University College London
Department Name: Computer Science


Understanding the functional makeup of the brain is a holy grail of neuroscience, and imaging tools play a significant role in achieve this aim. Functional imaging is a more difficult goal than anatomical imaging because it depends on establishing a contrast mechanism that relates to physiological function and also can be measured with good accuracy and resolution. Optical techniques are very attractive because of the rich information encoded in the absorption spectra of many different molecules, but they are difficult to use at large scale because of the high degree of scattering that occurs in passing through different tissues of the body.

By using time-resolved measurements of the propagation of light from multiple illumination patterns, diffuse optical tomography (DOT) can produced low-resolution images of absorption and scattering properties, and decorrelate these to produce maps of oxygenation in the brain and other organs. At the same time, diffuse correlation spectroscopy (DCS) examines the way in which coherent light is decorrelated from itself when compared over time. This decorrelation naturally occurs due to the Brownian motion of endogenous scattering particles, and blood flow. Coherent optical techniques thus allow the non-invasive monitoring of blood flow and provide an indication of pathological cerebral auto-regulation during, e.g., stroke. Until recently, limitations in coherent detection technology have prevented significant developments towards diffuse correlation tomography (DCT), wherein volumetric images of blood flow can be produced.

In this project we aim to develop a system for DCT and time-resolved DOT in one device. This will bring the two techniques together to provide images of cerebral blood flow and cerebral metabolic rate of oxygen extraction in the brain for the first time. We propose the development of theoretical and experimental methods which will enable the development of a new generation of optical instruments for portable, low-cost, continuous simultaneous monitoring of blood flow and chromophore concentrations. The rich images produced by our system have the potential to vastly improve our understanding of underlying neurological processes and pathology, and to allow the efficient use of scarce resources in targeted treatments.

Planned Impact

Impact in Neuroscience
In combining standard diffuse optical methods with correlation tomography, we introduce two new measurement parameters, namely the cerebral blood flow (CBF) and cerebral metabolic rate of oxygen extraction (CMRO2), which will enable a better understanding of the physiological status of the brain in a complete, accurate, and robust way. The combination of these two techniques will be unique worldwide. The value of the flow measurement is essential, owing to the link between the blood flow and the oxygen consumption: if a cell is using more oxygen it will demand more blood flow. By measuring both, we will be able to determine if the tissue is actually using more oxygen, indicated by an increase in flow, and a decrease in oxygen saturation. In so doing we expect to increase the understanding of the basic functional connectivity of the developing brain which will impact on other areas of cognitive neuroscience.

Impact in Healthcare
According to the global Action Report published by the World Health Organisation in 2012, preterm births are 15 million every year and rising. More than 80% of preterm births occur between 32-37 weeks of gestation and most of these babies can survive with essential newborn care. More than 75% of deaths of preterm births can be prevented without intensive care. Even though the risk of death to preterm infants is reducing, the ongoing potential for brain injury leads to severe disability, both physical and cognitive.
Through enabling the continuous monitoring of babies with existing or suspected impairment, our technology will also allow the targeted use of scarce rehabilitatory resources.
The long term impact of this research will be to improve the knowledge of pathophysiology of brain injury in premature babies in order to reduce the risk of brain impairment.

Impact in Imaging Science
Recent trends in imaging science have also seen the benefit of combining two or more imaging modalities, either from different scanners or on the same system. A well-established method to combine the information from different modalities is to merge them on the image level, i.e. the images are first reconstructed and then the information is combined (image fusion). However, when one or more of the modalities is ill-posed, as is the case here, then image reconstruction should properly be treated as the recovery of joint information simultaneously and entails the proper handling of joint prior information. We expect that the techniques developed in this project will produce impact in other multimodality/multispectral methodologies such as PET-MRI, MRI-EIT, and SPECT-CT, amongst others.

Impact in Data Science
We will develop novel methods for handling big datasets: three dimensions of space, two of time (short correlation time, long physiological time) and multispectrally. This "6D" problem will lead to novelty in compression methods and algorithmic design can be expected to impact on other areas of data science where continuous modelling of long time-series and large data arrays is relevent; notably geosciences and geostatistics, astrophysics, and the communications industry.


10 25 50
Description 1. We have developed dual-wavelength spread spectrum-based time-resolved system for functional NearInfrared spectroscopy applications. We were able to monitor functional responses from adult subjects using a spectroscopic, VCSEL-SPAD-based TD system. The instrument was capable of detecting small haemodynamic changes and also to provide repeatable estimates of relevant time-domain information, in a low-noise manner.
2. We developed a novel deterministic solution strategy for the (time-dependent) Radiative Transfer Equation (RTE) using pseudo-spectral time-domain method and spherical harmonic expansion in angle. A combination of perfectly matching and perfectly transmitting layers allowed the modelling of different boundary conditions.
3. We developed an efficient adjoint Monte Carlo method for inverse problems in optical tomographic modalities.
Exploitation Route The low cost and small footprint of this novel time-domain approach to fNIRS suggests potential compared to traditional TD methods and opens up several promising future developments.
We believe that miniaturised, low-cost, VCSEL-SPAD instruments could revolutionise the way contemporary TD diffuse optics systems operate and become the next-generation instruments for comprehensive human brain studies.
The new solution strategies using pseudo-spectral time-domain methods can have impact in the communications industry for e.g. underwater communication through random media
Sectors Digital/Communication/Information Technologies (including Software),Healthcare

Description Time-domain fNIRS is an increasingly active field with commercial developments appearing worldwide. The development of stochastic reconstruction methods has begun to have wider impact. We are beginning to develop these methods in other inverse problems including in conjunction with industrial partners. In particular we have developed related methods for Positron Emission Tomography (PET) which show considerable speed up compared to other stochastic iterative methods.
First Year Of Impact 2021
Sector Healthcare
Impact Types Economic

Description Stochastic iterative regularization: theory, algorithms and applications
Amount £385,265 (GBP)
Funding ID EP/T000864/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 08/2019 
End 08/2022