CAMIRA: Correction Advances for Myocardial Imaging using Registration Algorithms

Introduction: Myocardial perfusion imaging uses a gamma camera to image the perfusion of the heart to evaluate the presence of coronary artery disease. The Discovery NM 530c dedicated solid state cardiac gamma camera (DNM 530c) is an inherently three-dimensional imaging system that is different in design, and uses different acquisition parameters, compared to Anger gamma cameras. Aim: To determine the extent and frequency of patient and respiratory motion artefacts on myocardial perfusion images acquired on the DNM 530c and develop data driven motion estimation and correction techniques that utilise the three-dimensional nature of the system. Method: The effect of motion on myocardial perfusion images acquired on the DNM 530c was evaluated on phantom studies, and planar and three-dimensional techniques implementing image registration were developed for patient motion estimation. The technique was adapted to incorporate principal component analysis to facilitate the measurement of respiratory motion. Validation was performed on phantom simulations and explored through patient studies. Motion correction was applied by registering reconstructed binned data. Results: Patient motion ≥10mm that is present for ≥17% of the acquisition introduced significant motion artefacts. There was no significant difference (P=0.258) between the planar and three-dimensional patient motion estimation techniques. Motion correction removed artefacts from 9/10 phantom simulations. Cranio-caudal motion ≥8mm was measured on 10% of patient studies and 5% were affected by motion. No significant patient motion was identified in the lateral or ventral-dorsal directions. A strong correlation was demonstrated between the respiratory motion signal generated using the respiratory motion estimation technique and measured using an external device for two out of eight validation patients, with one patient demonstrating motion artefacts. Significant cranio-caudal respiratory motion was identified on 45% of patient images, with 4% demonstrating motion artefacts. Respiratory motion ≥15mm introduced artefacts. A quality index of ≥0.7 can be used to identify images that would benefit from motion correction; this would result in 1 in 3 patients undergoing correction. Conclusions: Data driven motion estimation techniques for both patient and respiratory motion on the DNM 530c have been developed. It has been demonstrated that patient motion ≥10mm that is present for ≥17% of the acquisition and respiratory motion ≥15mm can introduce artefacts into clinical scans. Acknowledgements: This report is independent research arising from a NIHR/CSO Healthcare Scientist Doctoral Research Fellowship supported by the National Institute for Health Research. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health