Joint cardiac and respiratory motion correction and super-resolution in coronary PET/CT.

摘要

Coronary artery diseases (CAD) such as atherosclerosis is a leading cause of mortality and morbidity in industrialized nations. Such diseases are marked by development of chronic vascular inflammation in coronary arteries, and accurate assessment and characterization of this inflammation through medical imaging methods is an important step towards the treatment of CAD. It has been shown that positron emission tomography (PET) is capable of detecting large vessel inflammation via activated macrophage uptake of a radiotracer such as [18F] -- FDG (Fluorine-18 -- Fluro deoxyglucose). However, respiratory and cardiac motions during image acquisition lead to severe blurring of the resulting images thereby rendering the spatial resolution inadequate for detection of the coronary arteries. The objective of this research is to develop novel algorithms that facilitate the ability to produce high resolution PET images of the coronary artery inflammation.;In this dissertation, we propose novel algorithms for performing motion aware reconstruction in PET which fall under two categories: image domain and projection domain methods. In the image domain category, we describe a novel method for joint cardiac and respiratory motion correction in PET/CT called Cardiac Shape Tracking with Adjustment for Respiration (CSTAR). It uses a sequential cardiac and respiratory motion correction scheme by decoupling the two motions, and also features the use of all acquired data for SNR preservation. CT images are used for cardiac shape tracking through the estimation of cardiac motion. Cardiac motion correction is incorporated in a super-resolution framework, followed by adjustment for the residual respiratory motion blur using blind deconvolution.;Later, we extend our previous image-domain approach to a fully integrated, data-domain method that starts from the observed projection data and performs a model-based inversion and motion correction of all the data to create a high-resolution focused cardiac image. We term the new approach Data-domain Cardiac Shape Tracking and Adjustment for Respiration or D-CSTAR. In contrast to existing image domain methods the image reconstruction and motion correction steps are not separated. Unlike current data domain methods both cardiac and respiratory motions are compensated for. In D-CSTAR, cardiac motion parameters are estimated from X-ray CT images acquired in a breath-hold state. This cardiac motion information is incorporated in a unified PET reconstruction functional which jointly estimates and corrects for respiratory motion, compensates for phase aligned cardiac motion, and super-resolves the image. The technique is presented and applied to simulated cardiac PET/CT data corresponding to the XCAT phantom with both cardiac and respiratory cycles. The results show a marked qualitative and quantitative improvement when compared to conventional and existing PET methods.