Wearable devices offer a wealth of data on users' physical activity, sleep patterns, heart rate, and other physiological signals. Use of such devices could lead to tangible health benefits (such as earlier diagnosis, better prediction, or personalised treatments) by aiding in the evaluation and management of various conditions through continuous real-world measurements.

As part of the Imaging and Engineering Smart Devices and Sensors (wearables) subtheme’s work we are establishing a lab testing capability for consumer and medical smart devices and sensors. Wrist worn sensors such as FitBit provide useful information about a wearer’s health through measurement of their physical activity (step count) and vital signs (heart rate). Meanwhile, smartphone based health monitoring applications enable data collection through a user’s personal devices. 

Google Health and FitBit are partners in the South Yorkshire Digital Health Hub and can provide researchers across the BRC, commercially available devices to be used within their studies. Prior to widespread incorporation into clinical research studies within the Sheffield BRC, we wish to conduct a validation of the metrics provided by these applications and devices to ensure their precision and fitness for purpose for clinical and patient users. 

The proposed study which will be conducted with a healthy population of research participants will complete a number of structured mobility tasks in the laboratory and in the real world.

The student will assist the research team in conducting a research study:

• recruitment of healthy control participants and scheduling of assessments

• conducting technical validation data collection including a gold standard measurement system for heart rate and step count

• analysis of performance of device outcomes including measures of accuracy, and reliability 

University of Sheffield Research Ethics committee approval is already in place for pilot assessments to take place at our Motion Capture Gait Laboratory located within the Mechanical Engineering Department.

Identifying Alzheimer’s Disease and related forms of dementia in stages before the clinical symptoms are observable is crucial in preventative and early identification efforts. Previous research has shown that middle-aged individuals with risk factors for Alzheimer’s Disease show patterns of both structural and functional brain changes similar to what is observed in Alzheimer’s Disease. 

To delineate the different stages of pre-symptomatic Alzheimer’s Disease, we aim to outline the fundamentals of the effects of risk factors on the resting-state functional connectivity, and their interactions with other Alzheimer’s Disease related pathology, and demographics. In doing so, we aim to compare the differences in the connectivity in major resting-state networks between the middle-aged individuals with and without risk factors. We will use a candidate-gene approach and control for demographic, environmental, and medical factors. This project is part of a longitudinal study within the international research initiative PREVENT-Dementia, which will provide a novel knowledge of the trajectory of Alzheimer’s Disease-related vulnerabilities in the brain throughout the lifespan. This project will complement our previously published and ongoing work on the effect of multiple genetic risk factors (i.e., APOE, MAPT, CD33, CLU, ABCA7) on functional connectivity and its relationship with structural changes and cognition across the lifespan.

Our research group is closely aligned with the Sheffield Ataxia Centre. Ataxia is a very debilitating condition where maintaining balance and co-ordinated movements become increasingly worse, such that people end up having not being able to walk and use a wheelchair and may have increasing difficulty in feeding themselves or getting dressed by themselves. One of the main reasons for it are diseases that affect a part of the brain called the cerebellum. Sheffield is one of the largest ataxia centres in the world. The Sheffield Ataxia Centre is a specialist medical clinic based at the Royal Hallamshire Hospital, which treats patients with cerebellar ataxia; in this condition the part of the brain called the cerebellum undergoes often very severe atrophy (shrinkage) and over time becomes “withered” in appearance. 

In researching cerebellar ataxia a relatively basic methodological need is to “segment” the cerebellum from a brain MRI scan, i.e. to “cut it out” from the rest of the image, in order to calculate its volume (this being an important variable to determine if atrophy has occurred and by how much). There exist many pieces of software which attempt to do this automatically and these often work very well on healthy controls or people with only mild cerebellar disease. However, they tend to fail the more deformed or shrunken a cerebellum becomes. 

We have recently completed a large project which includes brain MRI scans of patients with ataxia and have run into this problem. The segmented results frequently fail in instances where the cerebellum is highly atrophied by either including too much or too little. We believe that an AI approach may be far more accurate and wish to collect some pilot data to support doing this.

In this project, the student would be trained to learn the anatomy of the cerebellum and then take the segmentations already made and manually correct them. This wouldn’t be done across the whole dataset but on some targeted subgroups of patients defined by how severe their atrophy is. By creating “gold standard” cerebellar masks in this way and comparing them to the original automated outputs, we would look to first show quantitatively that the mis-segmentation does indeed get worse the more atrophy has occurred. Secondly, the manual masks could then be used in training an AI approach to ultimately improve this method.

The production of an accurate cerebellar segmentation routine would greatly increase our ability to run high quality research studies on these patients, and ultimately better understand the condition and provide care.

The project will analyse data obtained using IMU sensors from a performance test called the timed up and go. 

This test consists of a sit to stand from a chair, walking a set distance, turning 180 degrees, walking back to the chair and sitting back in it. This test is a standard assessment in elderly and Neurological patient cohorts. Typically this uses time to complete the task to interpret the impairment the patient is facing, however this methodology loses quite a lot of potentially useful information which may be obtained through the test such as time taken to stand, time taken to turn and number of steps taken during the turn. All of these aspects may provide pertinent information which may inform clinicians about the impact of a disease, and the impact of drug therapies.

We are seeking to improve an existing algorithm to better recognise transitions during the timed up and go test. This project will involve data processing in MatLab and interpretation of IMU data obtained in a variety of Neurological diseases and healthy cohorts.

Heterogeneity in hyperpolarised-gas MR images of the lungs is typically summarised with just one quantity: the Ventilation Heterogeneity Index (VHI), computer as the interquartile range of the Coefficient of Variation (CoV) map, or the median of the CoV map.

Investigating the actual distributions of the CoV maps across a large dataset of know clinical diagnosis harbours the possibility of finding additional biomarkers derived from those values. The project aims at testing this possibility, first by exploring all the different distributions and assessing the performance of biomarker candidates; then by validating the potential biomarkers, discovered during the first stage, against a healthy volunteer cohort by assessing the statistical significance of their comparisons. A final report will summarise materials, methods, and outcomes.