Over the past few decades, the use of ex vivo MRI has become widespread. This phenomenon was largely driven by the early development of various mammalian ‘brain atlases’ for neuroscientific applications as well as the need to characterize metabolism and other pathways in cells, isolated organs and cancer models. These developments included pioneering work that resulted in the creation of the first public, 3D, ex vivo atlases of the mouse (Benveniste et al. 2000, Dhenain et al. 2001, MacKenzie-Graham et al. 2003, Zhang et al. 2006) and rat brain (Veraart et al. 2011) for neuroscientific and translational applications (Benveniste et al. 2002). Moreover, these studies were the antecedents of recent big-data projects such as the Allen Brain Atlas (Lein et al. 2007) and the Mouse Connectome project (Oh et al. 2014). In other landmark studies, ex vivo MRI was employed to interrogate the biophysics of various cell types (Aguayo et al. 1986, Ackerstaff et al. 2001), isolated organs and solid tumors (Aguayo et al. 1987, Chatham et al. 1991) under a range of physiological conditions using an array of preclinical disease models. Collectively, these early studies set the stage for more unconventional applications of ex vivo MRI. Recent advances in MRI hardware, RF coil design, pulse sequence design (Badea et al. 2012), image processing and visualization software (Walter et al. 2010), the availability of complementary modalities such as optical and micro-CT imaging (Kim et al. 2012), and more affordable computational power have driven a slew of new applications of ex vivo MRI. Therefore, recent applications of ex vivo MRI that are ‘off the beaten path’, or ‘beyond rodents’ are the focus of this lecture.

For example, Deng et al elegantly demonstrated how the integration of imaging data and computational modeling could facilitate the development of more accurate models of infarct-related arrhythmic circuits (Deng et al. 2015). In this study, the authors developed computer models based on ex vivo MRI images of pig hearts with ischemic cardiomyopathy and showed that they could predict and analyze ventricular tachycardia resulting from a specific infarct architecture. In the future, such image-based models could assist in clinical decision making and ablating reentrant cardiac pathways. Another example of this integrated modeling and imaging paradigm comes from the work of Huang et al in which ex vivo MRI data of coronary plaques from patients were used to construct 3D fluid-structure interaction computational models (Huang et al. 2014). These models were used to investigate the association between plaque wall stress and coronary artery disease (CAD). They found that plaques from patients who died from CAD were associated with higher critical plaque wall stress compared to patients who died from non-CAD causes.

Another application of ‘mesoscopic’ or intermediate resolution ex vivo MRI is its utility as an ‘integrator’ or ‘bridge’ modality between imaging data acquired at the macroscopic and microscopic spatial scales. Recently, Cebulla et al demonstrated the feasibility of 'multiscale' angiogenesis imaging in a human breast cancer model, wherein they bridged the resolution gap between ex vivo micro-CT and in vivo MRI using intermediate resolution ex vivo MRI (Cebulla et al. 2014). They showed the feasibility of creating co-registered maps of vascular volume from three independent imaging modalities, and were also able to visualize differences in tumor vasculature between viable and necrotic tumor regions by integrating micro-CT vascular data with tumor cellularity data obtained using diffusion-weighted MRI.

Ex vivo MRI is particularly useful for correlating post-mortem human tissue samples with histopathology. This is an especially powerful paradigm for characterizing rare disorders or disorders for which high-quality in vivo imaging data are scarce. One recent example employed ex vivo MRI to characterize ischemic cavities in the cerebella of ageing subjects (De Cocker et al. 2014). Histopathological correlation allowed them to classify these cavities more accurately and gain insight into the imaging characteristics of these cavities so that in the future they can use this information to reliably identify patients with subtle manifestations of cerebrovascular disease in the cerebellum. Another area where ex vivo MRI can play a crucial role is in the imaging of small structures that change rapidly over time. An elegant example of this comes from Wang et al, in which they used ex vivo MRI to obtain high-quality images of the detailed local neuroanatomy in early fetal stages because the fetal brain is very small with a complex architecture that rapidly changes during brain development (Wang et al. 2015). They were able to identify the four-layer-structure within the fetal cerebral wall as early as 10 gestational weeks, and five to six layers during the early second trimester. More recently, several studies have harnessed the strengths of high-resolution ex vivo diffusion tensor imaging to map the 3D complexity of the white matter fiber architecture in the human brain. These studies include the human cerebellum (Dell'Acqua et al. 2013), brainstem (Aggarwal et al. 2013) and cortex (Aggarwal et al. 2015).

A recent study demonstrated the utility of ex vivo MRI as a potential new clinical tool for breast cancer detection (Agresti et al. 2013). In this study, ex vivo MRI of resected breast lesions was conducted and helped detect malignant tumors within several of the surgical specimens. This helped physicians verify that the initial tumor excision was successful and that the disease was contained in the surgical specimen. It also demonstrated that conventional imaging underestimated the extent of certain breast tumor subtypes. Collectively, these data demonstrate that ex vivo MRI helps with tumor staging and ensures free margins of resection, while simultaneously helping to avoid surgical re-excisions and additional unplanned surgery. Luciani et al conducted ex vivo MRI on excised axillary lymph nodes from breast cancer patients (Luciani et al. 2009). They were able to precisely correlate findings from conventional imaging and pathology, as well as identify nodal features that were suggestive of metastatic involvement. Orczyk et al used ex vivo MRI to quantify alterations in the prostate resulting from surgical excision (Orczyk et al. 2014). They systematically compared in vivo and ex vivo MRI data from unfixed freshly excised specimens and found that the observed prostate volume was significantly smaller on ex vivo MRI than in vivo MRI. This was likely due to loss of vascularity and connective tissue following excision. Moreover, these findings have profound implications for clinical co-registration systems that are currently in development for pathological validation and patient follow up (Goubran et al. 2015). Along similar lines, ex vivo MRI was recently used to generate 3D models of brains and then rapid prototype 3D brain holders (Guy et al. 2016). These customized 3D-printed brain holders maintained brain positioning between ultra-high resolution MRI and tissue cutting for histology. The preservation of brain position enabled accurate comparisons between ex vivo MRI and histology as well as to in vivo MRI. Finally, in another ‘beyond rodent’ application Xiao et al used ex vivo MRI to monitor the stability and healing of lumbar intervertebral disc transplants in goats (Xiao et al. 2015). Post-mortem images demonstrated that the disc allograft had united well with the host vertebral bones and was still hydrous at 3-month follow-up.

To summarize, while ‘traditional’ applications of ex vivo MRI for the development of mammalian brain atlases (e.g. marmosets, squirrels etc.) continue , unconventional applications ranging from image-based modeling, to multiscale imaging, clinical applications, rapid protoyping and transplantation also continue to be developed. Collectively, these advances herald an exciting new era for ex vivo imaging unlimited by the traditional constraints of hardware and software. More importantly, ex vivo imaging is slowly beginning to inform the clinical workflow for patients with different types of cancer, and other debilitating disorders.