Animal imaging using L1-regularized Quantitative Susceptibility Mapping

Introduction: Quantitative susceptibility mapping (QSM) is a technique that uses phase data from an MRI image to estimate the susceptibility distribution in the object. It has been demonstrated that QSM is able to correctly estimate the magnetic moment of specimen differing in susceptibility to the surrounding tissue [1]. We would like to exploit this ability to perform quantitative imaging of biomarkers in animal imaging. However, animal imaging presents additional challenges: the need for higher resolution suggests lower SNR; mixes of several tissues can create significant artifacts that impede quantification. In this work, we estimated the susceptibility change induced by SPIO nanoparticles that are targeted to specific cells. In experiment (1), we scan a rat brain after stroke injected with neural progenitor cells (NPCs) incubated in a solution containing a suspension of ferumoxide-protamine sulfate [2]. In experiment (2), we image a mouse injected with SPIO nanoparticles that target the intercellular adhesion molecule ICAM-1, which is induced in response to inflammation. We use total-variation based regularization to circumvent the problems with low SNR and the streaking artifacts. Materials and Methods: In both experiments, we used a Bruker 7T scanner to obtain high-resolution 3-D T2* images. In experiment (1), the resolution is 0.1μm isotropic. Prussian blue staining was performed for reference in SPIO location. In experiment (2), the MRI resolution was 0.2 μm isotropic. To have reference images, we used the nanoparticles coated with the Alexa Fluor 488 labeled-HA I domain (F265S/F292G), which allows use of fluorescence microscopy to confirm the location of SPIO. The computations were performed in MATLAB.