Investigation of Personalised Post-Reconstruction Positron Range Correction in 68Ga Positron Emission Tomography Imaging

Abstract

Positron range limits the spatial resolution of Positron Emission Tomography, reducing image quality and accuracy. This thesis investigated factors affecting the magnitude of positron range, developed a personalised approach to range correction, and demonstrated the approach using simulated, phantom and patient data. The Geant4 Application for Emission Tomography software was utilised to model positron range when emitted by radionuclides, namely 18F and 68Ga, in water, bone and lung. The impact of range blurring in lungs was found to be ten times larger than in bone and four times larger than in water or soft tissue, regardless of the positron energy. Range effects occurring with different isotopes (18F and 68Ga) were evaluated across measurement and reconstructed spatial resolutions. It was found that range correction was not necessary when using 18F for voxel sizes larger than 4 mm. In contrast, range correction was required for images generated using 68Ga, particularly within or adjacent to the lung. An iterative, post-reconstruction range correction method was developed which relied only on the measured data. The correction method was validated in both simulation and phantom studies. Image quality and quantification accuracy of corrected images was shown to be superior when imaging with 68Ga. Importantly, the range correction suppressed and controlled image noise at high iteration numbers. Finally, in a patient study, image noise in regions of uniform uptake was significantly increased by ~2% (p<0.05), yet mean standardised uptake values remained unchanged after correction, showing the same uptake for normal radionuclide distributions. The lesion contrast and maximum uptake values were improved by 20% and 45%, respectively with statistical significance (p<0.05). Although these promising results show that the proposed method of range correction can be generalised to reconstructed images regardless of measurement system, acquisition parameters and radionuclides used, further research is warranted to improve the method, particularly with respect to removing or reducing the artefacts which were shown to impacted reader preference.