Daniel West is a PhD student student Integrating myelin imaging with diffusion MRI for microstructure modelling with Dr. Shaihan Malik and Dr. Jacques-Donald Tournier and is based at King’s College London.
Myelin is a substance that forms an insulating layer around nerve cells in the brain, playing an important role in normal brain function. Damage to myelin hinders transmission of signals and can have serious neurological consequences. Links to diseases such as Alzheimer’s and Parkinson’s are well-established. However, due to a compact molecular structure, it is difficult to image myelin using conventional magnetic resonance imaging (MRI) techniques and so it is ignored in existing microstructural models. More sophisticated methods are currently under development and promising clinical images have been obtained but are widely disputed by statistical analysis studies. The aims of my project are to investigate physical mechanisms underpinning common myelin imaging techniques through computational simulation and using this information, either address their limitations or design a new, more specific method.
MRI is sensitive to diseases in white matter (WM) and is commonly used to diagnose lesions and monitor treatment of multiple sclerosis, for instance. Lesions appear hyperintense (bright) in an image but this can also be caused by many other inflammatory and non-inflammatory diseases, so diagnosis can be complicated. Successful imaging of myelin (a major component of WM) remains one of the big unanswered questions in MRI, and an attempt to develop a consistent and mathematically-robust technique could represent a considerable step forward in brain imaging. The recent discovery of novel therapies for neurological diseases makes early diagnosis and treatment imperative to ensure an improved survival rate. This may be achieved using a more sensitive myelin imaging method capable of identifying neurodevelopmental abnormalities and neurodegeneration.
During completion of an MSci Physics degree at King’s College London, I became interested in medical applications. Research projects entitled: “Fluorescence Microscopy of Cell Membranes”, “Interaction of PRODAN with Membrane Systems” and “Hyperspectral Micro-Imaging for Cancer Tissue Classification” enabled me to develop a wide-range of skills applicable to medical imaging, from mechanical engineering to computational modelling. Having studied nuclear magnetic resonance, I became fascinated by MRI and am excited to explore its unique potential for neurological disease diagnosis, especially in the developing, infant brain. My PhD project enables me to combine specialisms in biophysics and medical physics alongside prominent, multi-disciplinary researchers at St Thomas’, as part of the CDT and Perinatal Imaging Department. The non-existence of clinically-established techniques to image myelin, given its prominence in neurological complications, surprised and motivates me to contribute a solution to a real-world medical problem.
To date, my work has been solely computational-based but research scanners at St Thomas’ will soon be used for phantom and in vivo investigations. When introduced to a new concept, my first ‘port-of-call’ is a literature database, where I can enter search criteria to locate all relevant publications. I then collate these in a reference manager and begin reading to slowly build-up my knowledge. Since most of my work is mathematical, I often find that a pen and paper are the most useful and efficient tools when working through a derivation and attempting to tie together different analytical approaches used in the literature. Currently, every simulation I have run has been coded in MATLAB due to its suitability for imaging-based problems and extensive documentation. MRI sequence-specific functions have been written that solve a particular type of first-order differential equation (known as Bloch-McConnell) using a matrix exponential method. I have parallelised my code and extensively used MATLAB’s optimisation toolbox that determines parameters for minimisation of objective functions under certain constraints. To communicate my findings, I tend to accumulate data plots and small snippets of analysis in Evernote for presentation in group meetings and departmental journal clubs.
Having performed a quantification of MT-induced parameter bias in mcDESPOT, the focus now shifts to testing existing correction strategies (such as CSMT) and exploring alternative signal fitting algorithms. This can be achieved using simulated data and graphical representation of parameter search-space but acquired experimental data will be used to corroborate findings. CSMT should provide a superior parameter estimate accuracy and precision, and demonstration of this may represent a novel finding in the context of mcDESPOT and MT. My simulations are all completed using a so-called Bloch isochromat-based framework, but another method exists: extended phase graphs (EPG). It may be beneficial to establish these simulation tools to replicate any discoveries. Subsequent phantom experiments can be completed, but as is often the case in scientific research, my project may take an alternative path. mcDESPOT has considerable clinical potential but its mathematical flaws could lead to investigation of a completely different imaging technique!