Early Diagnosis, Prognosis and Treatment
What’s going to happen over time? How can we change what will happen?
Recognising what has gone wrong (diagnosis) and predicting what will happen over time (prognosis) is essential for guiding appropriate and timely treatments and interventions.
Our neuroimaging research takes targeted approaches to improve clinical applications for patients with neurological and psychological conditions:
Finding hidden markers to enable early diagnosis
By optimising our neuroimaging measurements, we can see abnormalities in neuroanatomical microstructures and brain activity that are not visible to the human eye. These abnormalities can sometimes be detected before other disease-defining symptoms have emerged.
Building and mining datasets to enable more accurate prognoses
By combining data from neuroimaging and behavioural tests that are acquired over multiple time points, we can understand how symptoms change over time, in different conditions. This knowledge can then be used to predict recovery or decline in new patients with the same disorder.
Outcome predictions can be optimized by using machine learning to combine many different types of data that might influence the outcome (i.e. demographic, behavioural, cognitive etc). Prognoses can be generated by testing how accurately predictions from one cohort of patients predicts outcome in another cohort of patients.
Providing feedback to enable more effective treatments
By identifying information about the actual or predicted effects of treatment, neuroimaging can track progress, guide recommendations, and even be part of the treatment itself. For example, we are using neuroimaging to:
- Develop novel neuroimaging-based treatments that use neurofeedback during behavioural tasks to focus attention on the most effective recovery strategies.
- Understand how neuroanatomy changes with behavioural and pharmaceutical interventions that slow or reverse disease progression.
- Provide data analysis tools for determining the optimum dose of treatment.
- Select behavioural tasks that probe the function of regions targeted for neurosurgery during intra-operative mapping. If a brain region responds, the surgeon may preserve the region to avoid post-operative loss of essential human functions.
Early diagnosis will improve long-term outcomes
Recognising a disease early (i.e. diagnosis) means that patients can understand how their symptoms will change over time (prognosis) and how this can be changed with treatment. Earlier diagnoses will also allow targeted interventions to be commenced earlier, before potentially irreversible damage to brain tissue has occurred, helping to slow or reverse the disease process.
Safe and accessible diagnoses
Neuroimaging provides non-invasive diagnostic tools that can reduce the need for invasive technologies, and therefore make some diagnostic tests more widely accessible across healthcare systems, more comfortable for patients, and associated with lower risks of adverse events.
Accurate prognosis improves treatment options
The anticipated rate of clinical clinical change (i.e. prognosis) is important to help direct the right treatment to the right patient. Individuals with the same condition, but different prognoses, are likely to require different types of interventions. Prognoses can also be used to evaluate new treatments by providing a benchmark for testing how much the treatment alters the expected course of recovery or decline.
DIAGNOSING HUNTINGTON’S DISEASE
Huntington’s disease is characterised by uncontrolled movements, emotional problems, and loss of thinking ability (cognition). Work at the Centre by Sarah Tabrizi and Geraint Rees’s team includes developing and testing a model in Huntington’s disease which will allow disease progression to be monitored years before the diagnosis in those at risk of developing Huntington’s disease (Gregory et al., 2018).
Parkinson’s disease is a common progressive disorder of the brain that becomes more likely with age. The main symptoms include tremor, slow movements and stiff inflexible muscles. However, the way the disease affects an individual, from the initial symptoms through to how it progresses over time, is highly variable. Once individuals begin to experience movement problems, ~70-80% of the dopamine producing nerve cells within a brain region called the substantia nigra have already been lost.
At the Centre we are trying to understand why Parkinson’s disease is so variable, researching how to accurately diagnose the condition earlier, and develop ways to predict how it will progress over time. Achieving these will help develop future treatments that can be tailored to individual subjects and started before significant loss of brain tissue:
Using high-resolution quantitative MRI, deep clinical phenotyping, genetics and serum biomarkers, Christian Lambert’s team aim to understand how individual differences in brain structure are related to the clinical variability in Parkinson’s disease. Through this, the aim is to develop ways to predict how quickly the disease will progress, and allow accurate diagnosis before movement problems develop. Click here to find out more about the longitudinal qMAP-PD study.
Rimona Weil’s team are using neuroimaging to identify which patients with Parkinson’s disease are at risk of developing dementia.
PREDICTING CHANGES IN MOOD
Through assessment platforms on mobile apps, Ray Dolan’s team have demonstrated that it is possible to predict the subject’s mood fluctuations over time. This research could lead to predictions of relapse in people with clinical depression (Eldar et al. 2018)
PREDICTING OUTCOME AFTER STROKE
Cathy Price’s team has developed a web-based system that uploads MRI or CT scans of patients with speech production difficulties after stroke and outputs a prediction for whether and when speech will recover. The prediction is primarily based on identifying which parts of the brain have been damaged in the new patient and how the same type of damage affected the outcome in other patients (from our database) who have already been extensively studied. The database currently includes anatomical brain imaging and behavioural assessments acquired from multiple time points in more than 1200 stroke survivors.
In the future, the recovery trajectories generated for each stroke patient can also be used as a baseline measure to predict the effect of interventions on speech production (Neurotherapeutics Group – Alex Leff lab, Jenny Crinion lab).
The techniques used can also be applied to a range of other disorders (e.g. tumour or lesions caused by neurosurgical resections) and impairments (e.g. movement and cognitive impairments (Nick Ward, Sven Bestmann) as well as predicting morbidity and hospital attendance (High-Dimensional Neurology – Parashkev Nachev, Geraint Rees).
Brain activity can be fed back in real time to patients performing the task. The patients can then attempt to control their brain activity by performing the task differently. Geraint Rees’s team have used this “neuro-feedback” training to induce neuroplasticity in regions of the brain where visual awareness and covert attention are processed (Ekanayake et al., 2018b). With Sarah Tabrizi, the techniques are also being used in patients with Huntington’s disease to alleviate motor symptoms.
In parallel, our physics team will continue to optimise and refine neurofeedback training through improvements in rapid data acquisition, image reconstruction, addressing participant motion during scanning.
RESTORING LEARNING PERFORMANCE
Pharmacological interventions can be better understood by using computational modelling to analyze brain activity. For instance, Ray Dolan’s team has found that administering L-Dopa reverses neural deficits and restored learning performance (Chowdhuri et al. 2013). This work contributes to improving symptoms for patients with depression, schizophrenia, anxiety and addiction.
REMOVING MALIGNANT BRAIN TUMOURS AND TREATING DRUG-RESISTANT EPILEPSY
Current pre-surgical assessments, using either non-invasive functional neuroimaging or deep brain electrodes, focus on a limited set of tasks (e.g. verbal fluency) that do not adequately pinpoint the functional contribution of the region(s) that may be dysfunctional after neurosurgery.
We design targeted assessments of functional neuroanatomy that can be used by neurosurgeons to plan the removal of malignant brain tumours or treat drug-resistant epilepsy.