Cerebral metabolic rate of oxygen (CMRO2) and oxygen extraction fraction (OEF) are valuable to investigate tissue viability and function. Numerous magnetic resonance imainge (MRI) methods have been proposed to estimate them quantitatively. While those methods are investigated in healthy subjects, a robust framework is elusive from literature for reliable CMRO2 and OEF estimation in pathological scenarios. This thesis developed algorithms that improve the accuracy, robustness and applicability of CMRO2 and OEF for both healthy and pathological subjects. First, quantitative susceptibility mapping (QSM)-based and quantitative blood oxygen-level dependent magnitude (qBOLD)-based CMRO2 method were combined to resolve the issues of the individual methods. Second, the cluster analysis of time evolution (CAT) was proposed to improve robustness of the combined model against noise. With the technical advances in this thesis, CMRO2 and OEF that are robust against noise and sensitive to pathological senarios, e.g. ischemic stroke lesion, can be estimated.