Clinical adoption of quantitative neuro-imaging AI based techniques

1) Brief description of the context of the research including potential impact

Numerous advancements have been made in recent years regarding the use of artificial intelligence (AI) in medical imaging. Nonetheless, very few AI tools make it into clinical practice. The primary goal of the current PhD project is to investigate key sources of error and develop uncertainty measures for quantitative neuroimaging AI tools. The focus will be on neuro-oncology, dementia, epilepsy and MS. Potential impacts of the project include the facilitation of clinical adoption of AI assisted volumetric analysis, AI enhanced acquisition times and the translation of uncertainty measures associated with quantitative MR measurements into the clinic.

2) Aims and Objectives

Aims: Development of methodology that will enhance clinical translation of AI based tools.
Objective:
1. Identification of key sources of errors in existing AI tools.
2. Development and validation of uncertainty measures for automated volumetric analyses.
3. Evaluation of the implications of using AI technologies in the speed up of acquisitions in neuroradiology.

3) Novelty of Research Methodology

As of now, many AI tools in the field of neuroimaging do not convey their degree of uncertainty concomitant to the prediction. Furthermore, it is not sufficiently well understood what the impact is of using AI to speed up the acquisition of MR images.

4) Alignment to EPSRC's strategies and research areas

This project aligns with the EPSRC's research aims within healthcare technologies. Automated extraction of information from clinical data or images is a high priority research area of the EPSRC's research on medical imaging.

5) Any companies or collaborators involved

Currently no companies or external collaborators involved.