Magnetic resonance imaging (MRI) is a widely used non-invasive imaging technique, with rich contrast for interrogation of tissue physiology and pathology. MRI can be used to examine hemodynamic conditions like perfusion and oxygenation qualitatively and quantitatively. This dissertation reports new techniques that employ dynamic imaging and quantitative susceptibility mapping (QSM) to overcome technical challenges for improved MRI perfusion imaging and oxygenation quantification. Perfusion imaging after gadolinium contrast agent administration is a common clinical practice, where dynamic MRI technique is used to track contrast bolus. It is desirable to have high temporal resolution to capture blood dynamics, as well as high spatial resolution to depict small lesions. Unfortunately, these requirements are limited by hardware and physiological conditions. In this dissertation, a dynamic MRI technique was realized, using fast spiral acquisition and a constrained image reconstruction algorithm, to achieve high temporal-spatial resolution for liver perfusion imaging. Tissue susceptibility provides unique contrast in MRI. Recent development of QSM technique has been applied in various clinical applications. In this dissertation, the dynamic imaging method was extended to multi-echo acquisition and combined with QSM to map gadolinium contrast agent concentration during the first passage for cerebral perfusion mapping. Blood oxygenation is determined by the amount of deoxyhemoglobin in red blood cell. Magnetic susceptibility of deoxyhemoglobin is a source of MR contrast for imaging oxygenation. This dissertation reports correction schemes in both data acquisition and image reconstruction of QSM for improved blood oxygenation quantification.