On conventional Single-energy Computed Tomography (SECT), lesions with an attenuation greater than 100 Hounsfield Units (HU) can be definitively diagnosed as calcification. However, low-density calcifications and hemorrhage may have overlapping attenuation ranges between 40 and 100 HU and, therefore, cannot be differentiated with SECT alone. On T2*-weighted Gradient Recalled Echo (GRE) MRI, these lesions appear as “foci of susceptibility” in which their signal is hypointense due to the magnetic susceptibility of the lesions differing from that of the background tissue.

Dual-energy Computed Tomography (DECT) and Phase-Sensitive Magnetic Resonance Imaging (PS-MRI) represent two new imaging paradigms which both have the potential to more accurately identify intracranial calcification and hemorrhage. In DECT, x-ray tomography is acquired at two tube voltages;​ because x-ray attenuation is energy- and material-dependent, the data can be used to differentiate between materials that may have the same signal level on SECT. PS-MRI utilizes the phase data from T2*-weighted MRI acquisitions to determine how the local magnetic field varies across the image. By applying post-processing algorithms such as Quantitative Susceptibility Mapping (QSM), the phase can be used to calculate the magnetic susceptibility of a lesion. Since calcifications are diamagnetic and hemorrhage paramagnetic, we can make inferences about a lesion’s composition from these algorithms.

The objective of this dissertation work was to characterize brain lesions, discovered with traditional imaging methods, as either hemorrhagic or calcific by using Dual-Energy Computed Tomography (DECT) and Phase-Sensitive Magnetic Resonance Imaging (PS-MRI). To this end, MRI-compatible phantoms featuring models of both calcific and hemorrhagic lesions were developed and validated. This resulted in two phantoms with biologically similar lesion models that were then used to test the feasibility of differentiating calcific and hemorrhagic lesions with PS-MRI post-processing methods, in which QSM was able to accurately differentiate calcific and hemorrhagic lesion models. Finally, we undertook a patient trial testing the feasibility of identifying calcification and chronic hemorrhage in humans using both DECT and QSM in which the two modalities had accuracies of 99.7% (327/328) and 99.4% (326/328), respectively. The two modalities were concordant for 99.3% (148/149) lesions with SECT attenuation under 100 HU.