| EPSRC Reference: |
EP/Y016009/1 |
| Title: |
Leveraging universal fractal geometry to develop new AI for neuroimaging |
| Principal Investigator: |
Wang, Dr Y |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Sch of Computing |
| Organisation: |
Newcastle University |
| Scheme: |
Standard Research - NR1 |
| Starts: |
02 October 2023 |
Ends: |
01 April 2025 |
Value (£): |
619,864
|
| EPSRC Research Topic Classifications: |
| Artificial Intelligence |
Medical Imaging |
|
| EPSRC Industrial Sector Classifications: |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
*The Unmet Healthcare Need*
Unmet healthcare need: In neurological conditions, the demand for non-invasive structural brain Magnetic Resonance Imaging (MRI) is rising, where most patients require at least one MRI, but visual inspection and reporting are currently a bottleneck. Current visual assessments usually scroll through 2D snapshots and can therefore also miss changes in the complex folded 3D shape of the human brain. We propose to develop new AI to quantify and report brain shape abnormalities in neurological MR images.
*The underpinning science*
Brain morphology, or the study of the shape and size of brain structures, is a particularly promising area for developing new AI for neuroimaging. Human perception is insensitive to subtle changes in complex shapes, such as the folded brain. One feature of such complex shapes is that they are often "fractal", meaning that relevant shape information is often distributed over a range of different spatial scales. We recently validated the universal fractal nature of the primate brain shape and demonstrated its agreement with predictions made based on a biophysical model of cortical folding. This allowed us to propose a novel re-conceptualisation of brain morphology measures as a function of spatial scale, and we demonstrated in proof-of-principle data that this approach reveals previously hidden effects.
*Our proposal*
We propose to develop new AI based on the universal fractal geometry of human brains. The AI will take unseen neuroimaging data from a new patient as input, update its own algorithmic parameters based on supplemental data and pre-trained knowledge to debias the input, and output an interpretable report on relevant brain shape abnormalities in the patient. Importantly, the universal fractal geometry provides a basis for debiasing and provides constraints for reliability and reproducibility testing. Furthermore, as our method is fundamentally rooted in geometry, the outputs include interpretable visualisations. This project will further support four early career researchers (three female) in an exciting area of AI in healthcare, and we will develop our AI closely with our network of clinical teams and stakeholders.
Our ultimate vision is to create an AI tool that seamlessly integrates into clinical workflows. Where previously multidisciplinary teams gathered to discuss treatment strategy for a given patient, we suggest that the AI also takes a seat at the table by providing reports and interactive visualisation to e.g. confirm pathological mechanisms, or highlight new information that was previously not considered based on visual inspection.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.ncl.ac.uk |