PhD Studentship: Hot spots and pressure points: Developing AI to identify normal and pathological periprosthetic osteoblastic activity for total knee arthroplasty on SPECT/CT

Project description:

Single photon emission computed tomography (SPECT-CT) imaging combines functional and anatomical data, enhancing diagnostics but remains underutilized and provides non-specific diagnosis. SPECT-CT’s clinical value in assessing painful knee replacements (TKRs) includes accurate 3D analysis of component positioning and tracer uptake. However, there is limited differentiation between expected and pathological radiopharmaceutical uptake equivocal results in many cases.

The aim of this PhD studentship is to develop an advanced artificial intelligence algorithm incorporating finite element analysis and computer vision technology to improve prediction of areas where increased radiopharmaceutical uptake is expected versus pathological uptake.

The team has a dataset of 45 SPECT/CT scans with TKRs and known patient outcomes.  These will be segmented for anatomical structures on CT and augmented with SPECT scans to provide anatomical and pathological ground truth images.  Current AI models of the knee anatomical structures derived from MRI, developed by Prof Ye, will be adapted using transfer learning or domain adaptation methods and mapped to FEA predictions of periprosthetic stress and strain areas. 

Training in machine learning, computer vision and finite element analysis, image interpretation of SPECT/CT and pathophysiology will be provided by expert supervisors in these fields.  

The Department of Health and Care Professions has strong links and a long working relationship with the orthopaedic team locally.  Images to support this study are already held by the group and software to facilitate the work is available.

Advert information