Low dose and quantitative myocardial perfusion imaging using multidetector row computed tomography

In spite of the fact that myocardial CT perfusion imaging was shown to provide quantitatively accurate estimates of myocardial perfusion since the 1980's, it is not routinely used in the clinic today. The work presented in this dissertation addressed two major technical challenges which have played a role in limiting its clinical acceptance, namely the presence of partial scan reconstruction (PSR) artifacts and concerns regarding radiation dose levels.;Two methods were developed to reduce PSR artifacts: the targeted spatial frequency filtration (TSFF) and temporal interpolation of partial scan sinograms (TIPS). TSFF requires a special acquisition mode in which the tube current is modulated to complete a sequence of full scans. The TSFF provides a robust method to effectively reduce PSR artifacts requiring only a small increase (14%) in the required radiation exposure, relative to a conventional partial scan reconstruction. TIPS provides an alternative for conventional myocardial CT perfusion scan modes where the modulation of the tube current is not available. It is particularly well suited for patients with relatively stable heart rates.;Two methods were developed to address the radiation dose challenge: the nonconvex prior image constrained compressed sensing (NCPICCS) and the time adaptive filter (TAF). NCPICCS holds great promise to dramatically reduce radiation dose in time-resolved CT imaging, by using only a limited number of angular x-ray projections. Unfortunately, its practical implementation would require the availability of pulsed x-rays, which are not available on current CT systems. Also, because NCPICCS is iterative in nature, it is relatively slow due to the computation time required. These limitations motivated the development of the non-iterative TAF method that is applied in the image space and requires relatively short processing times. TAF consistently reduced image noise, by at least twofold, while maintaining spatial resolution and fidelity of the perfusion estimates.;The TSFF and TAF methods were evaluated using animal (pig) experiments, in which myocardial perfusion estimates were derived using CT (F CT) and validated against estimates derived from microspheres (F_MSF). For typical physiological ranges of myocardial perfusion (0.5 to 4 ml/g/min) a linear relationship was found: FCT = 0.894•FMSF + 0.189, (R2=0.83, p<0.01). This thesis found that reproducible myocardial CT perfusion can be performed using radiation dose levels which are typically lower (2.0 to 8.0 mSv) than those of alternative techniques currently used in the clinical practice to measure myocardial perfusion, such as SPECT (10 to 30 mSv) and PET (5 to 15 mSv).