|Institution:||University of Washington|
|Keywords:||attenuation correction; CT; MBIR; PET; respiratory motion; ultra low dose; Medical imaging; Bioengineering; Bioengineering|
|Full text PDF:||http://hdl.handle.net/1773/40846|
Positron emission tomography (PET) is a commonly used imaging tool in the management of patients with lung cancer and is of considerable interest in quantitative imaging of the thorax. Mismatch of PET data with computed tomography (CT) attenuation correction (CTAC) due to respiratory motion is a known source of errors in PET imaging. In theory, this can be corrected by matching individual PET and CT phases which have been generated by respiratory-correlated PET and CT. However, due to the high variability of patient breathing patterns and the nature of the scanning time differences between PET and CT, current respiratory-gated CTAC protocols for the irregular breather may cause additional bias in the PET image values. A ten-fold extension of the CT scanning time duration helps reduce PET imaging bias, but leads to the higher radiation dose to the patient. Lowering the CT source flux level to reduce dose, however, leads to increased noise and bias. Here we test the possibility of using model based iterative reconstruction algorithms (MBIRs) for generating the sparse-view, ultra-low-dose (i.e. an order lower than current low-dose protocols) CTAC images for both phantom and patient PET data. We also propose a new variance estimation model, which considers statistical changes caused by the non-positivity correction process, for the MBIR algorithms. The model based iterative CT reconstruction approach does generate more accurate CTAC map compared to current approaches. However, since iterative reconstruction algorithms typically assume a normal distribution of the attenuation data, we tested if the assumption is still valid in the ultra-low-dose regime. The simulation and empirical ultra-low-dose CT studies showed a skewed post-log likelihood distribution in certain ranges. The information delineates the estimation limits of model based iterative reconstruction approach on the ultra-low-dose CT imaging, and potentially helps guide scanning protocols customized for a lowest-reasonable radiation dose.Advisors/Committee Members: Kinahan, Paul E (advisor).