ABSTRACT

Regenerative medicine has become a new therapeutic approach in which stem cells or genetically reprogrammed cells are delivered to diseased areas in the body with the intention that such multipotent cells will differentiate into healthy tissue and exchange damaged tissue. The success of such cell-based therapeutic approaches depends on precise dosing and delivery of the cells to the desired site in the human body. To determine the accuracy and efficacy of the therapy, tracking of the engrafted cells in an intact living organism is crucial. There is a great need for sensitive, noninvasive imaging methods, which would allow clinicians to monitor viability, migration dynamics, differentiation towards specific cell type, regeneration potential and integration of transplanted cells with host tissues for an optimal time period. Various in vivo tracking methods are currently used including: MRI (Magnetic Resonance Imaging), PET (Positron Emission Tomography), SPECT (Single Photon Emission Computer Tomography), optical imaging (OI), photoacoustic imaging (PAI) and ultrasound (US). In order to carry out the detection with each of the aforementioned techniques, the cells must be labeled either exogenously (ex vivo) or endogenously (in vivo). For tracking the administrated cells, scientists usually manipulate cells outside the living organism by incorporating imaging contrast agents (CAs) or reporter genes. Strategies for stem cell labeling using CAs will be reviewed in the light of various imaging techniques.

Keywords:In Vivo Imaging, MRI, PET, SPECT, SPIO, Cell Labeling

1. Introduction

Cell-based therapeutic approaches, utilizing stem cells and genetically reprogrammed cells, have a great potential for diagnostics of and treatment options for tumors [1] -[3] , cardiovascular [4] , neurological [5] -[7] and autoimmune diseases [8] or wounds [9] [10] , for which there are currently no effective medical solutions available. The distinguishing properties of stem cells are their self-renewal and high potency, which enable them to differrentiate into specialized cell types [11] . Cell-based therapies are part of so called personalized medicine and involve using therapeutic cells deriving from the patient (autologous transplantations), from donors (also other organisms) or immortalized cell lines.

The success of cell-based therapeutic approaches depends on precise dosing and delivery of the cells to the desired site in the human body. To determine the accuracy and efficacy of the therapy, tracking of the engrafted cells in an intact living organism is crucial. There is a great need for sensitive, non-invasive imaging methods, which would allow clinicians to monitor viability, migration dynamics, differentiation towards specific cell type, regeneration potential and integration of transplanted cells with host tissues for an optimal time period. Various in vivo tracking methods are currently used including: MRI (Magnetic Resonance Imaging), PET (Positron Emission Tomography), SPECT (Single Photon Emission Computer Tomography), optical imaging (OI), photoacoustic imaging (PAI) and ultrasound (US). Incorporation of an imaging contrast agent, specifically for a given imaging modality into humans is restricted by the Food and Drug Administration (FDA).

In order to carry out the detection with each of the aforementioned techniques, the cells must be labeled either exogenously (ex vivo) or endogenously (in vivo). For tracking the administrated cells, scientists usually manipulate cells outside the living organism by incorporating imaging contrast agents (CAs) or reporter genes. Strategies for stem cell labeling can be divided into two categories: direct labeling and indirect labeling. Indirect labeling (Figure 1, left) involves genetic modification of a cell-reporter genes are introduced into the cells and then translated into non-native fluorescent or bioluminescent proteins, enzymes or receptors which can be subsequently detected using one of the abovementioned modalities.

Direct labeling (Figure 1, right) does not require genetic interference and therefore is simpler and inexpensive. This method is mainly based on incubation of viable cells with CAs. Because of their similar sizes most of the nanoparticles used as various CAs can be taken up by the cells with the same endocytotic pathways as proteins, viruses and DNA. The endocytotic pathways include passive diffusion, phagocytosis, macropinocytosis, clathrinand caveolin-dependent endocytosis and clathrin/caveolin-independent endocytosis [12] .

2. Magnetic Resonance Imaging

Magnetic Resonance Imaging (MRI) allows non-invasive serial imaging for dynamic monitoring of cell migration in situ in high spatial resolution (<50 µm in physiological systems) or even imaging of the individual cells [13] [14] . One of the main advantages of MRI is the safety as it does not require exposition to any ionizing radiation, which enables repetitive imaging in the same living organism. MR scanners are common in clinics and laboratories and are used for performing standard anatomical and functional imaging of patients. MR scanners rely on Nuclear Magnetic Resonance (NMR) and in most cases measure the proton (¹H) response from water and other endogenous molecules. The strong homogenous magnetic field is generated and an additional brief radio wave is applied which results in resonance of (¹H) protons. After the radio wave source is turned off protons return to their resting states creating the signal measured and recorded by a scanner. Subsequently, a MR image is constructed based on the tissue-specific variables, e.g. spin density or relaxation time constants (called T₁, T₂ and T₂* relaxation times) [15] [16] . Conventional clinical MR scanners can be utilized in living objects for molecular MR imaging (visualization of subcellular structures and macromolecules, e.g. cellular proteins and lipids) and cellular MR imaging (visualization of whole cells) [17] . Cell MR imaging in living tissues and distinction between administrated and endogenous cells requires application of magnetic CAs (also called the contrast media) which, due to their magnetic properties, modify relaxation times. Various criteria have been used for classifying the MRI CAs according to their chemical composition, metal properties (paramagnetic or magnetic), route of administration (e.g. intravenous or oral), medical application, biodistribution and impact on the MR image [18] . There are also several features that should characterize a perfect CA for MR imaging studies [19] . First of all, it should be specific and sensitive but also biocompatible and non-toxic, as it must be able to transition from preclinical and biological interest (bench) into clinical use in humans (bedside). Long-term, serial and repetitive tracking should be possible which means a CA should neither dilute following cell divisions nor leak to the neighboring cells/tissues. Another important aspect is the stability in the circulatory system which prevents undesired uptake of the CA particles by macrophages. Despite the fact that increasing attention has been paid to

development of new advanced CAs and there are some promising candidates, at this moment there are unfortunately none fulfilling all of these conditions at once.

As already mentioned in the introduction, it is possible to label cells either in the living organism (in vivo) or outside of it (ex vivo). Endogenous labeling of cells relies on injection of the CA directly into the body. In order to make the labeling more specific, CAs can be enriched with ligands that target the specific cell type and/or facilitate internalization. The second strategy relies basically on incubating harvested cells of interest for an optimal time period in the medium containing the proper concentration of the chosen CA.

Magnetic CAs for direct labeling of the cells for cellular MR imaging are basically classified into two major groups based on their susceptibility to magnetic fields: paramagnetic and superparamagnetic CAs [20] . Paramagnetic materials or particles are magnetized in the presence of an induced magnetic field. When the external magnetic field is removed, the materials do not retain magnetic properties. Superparamagnetism is the extension of paramagnetism—superparamagnetic nanoparticles possess high magnetic susceptibility, which means that they are magnetized to a larger extent in moderate magnetic fields. Paramagnetic MRI contrast agents are called positive contrast media and they cause a reduction in the T₁ relaxation time, appearing bright on the MR images. Superparamagnetic MRI contrast agents on the other hand are called negative and appear dark on the MR images, they shorten T₂ and T₂* relaxation times [18] [20] [21] .

2.1. Gadolinium-Based MRI Contrast Agents

Gadopentetate dimeglumine (known as Magnevist^®) was the first MRI contrast agent approved by the FDA [22] . The main advantages of using gadolinium compounds as MR imaging tracers are the low background level in the human body and the fact that they have a low level of reuptake by phagocytotic cells and are therefore also less likely to create confounding false-positive images [23] . In order to facilitate the incorporation of paramagnetic contrast agents into the cells, transfection agents (TAs) have been applied, such as dendrimers, Lipofectamine, Lipofectin, calcium phosphate and poly-L-lysine [24] [25] . To bypass the cell membrane barrier, gadolinium compounds have also been coupled with cationic liposomes or membrane translocation signals/peptides, e.g. viral protein—HIV1 tat [26] -[30] . Other modifications of gadolinium chelates include adding groups that can form bridges with the cells’ exofacial protein thiols (EPTs). In this way, a Gd-containing compound is anchored to the cell surface and can be transported into the cytoplasm [31] .

However, there are concerns about the toxicity of gadolinium compounds for humans as gadolinium is an element that is not naturally present in the human body [32] [33] . Although as the MRI CA gadolinium is administrated in a chelated form, in lysosomes (low pH) dechelation may occur and lead to generation of free Gd³⁺ ions. Free gadolinium ions are inorganic calcium channel blockers and are considered highly toxic due to their interference with the calcium-channel dependent processes and with the activity of some enzymes [32] [33] . Nefrogenic systemic fibrosis (NSF) is a rare but potentially harmful clinical syndrome occurring in kidney failure patients after exposure to Gd-containing contrast agents. The exact pathogenic mechanism underlying this condition remains unknown but it may occur months after the gadolinium CAs have been administrated to the patient. The cases of NSF are limited to individuals with renal failure as kidney disorders prevent gadolinium compounds from being removed from the body [33] . However, studies show that exposure of healthy skin to gadolinium-based MRI contrast agents induces changes that may lead to fibrotic disease [34] .

2.2. Manganese-Based MRI Contrast Agents

Manganese is also a very efficient positive contrast agent for MRI studies and unlike gadolinium is a natural cellular component as a cofactor for some key enzymes and receptors. Nevertheless, toxicity concerns exist here as well, as high concentrations of Mn²⁺ are neurotoxic and lead to manganism, which in many ways resembles the Parkinson disease. Manganese-based contrast media can be divided into two broad groups: small molecular agents like manganese (II) chloride; and nanoand macromolecular agents like hydrophobically modified MnSPIO or PEG-coated manganese oxides [35] [36] .

A manganese salt (MnCl_2; approved by FDA as a CA) has been used to label human embryonic stem cells (hESCs) and human bone marrow stromal cells (hBMSCs) [37] . In this study, cells were labeled directly for 30 - 60 min and then successfully tracked by MR imaging in vitro and in vivo. Importantly, as MnCl₂ in the living cells is transported by calcium channels, the signal on MR images comes only from viable and biologically functioning cells [37] . In another study [38] , the double labeling was performed; rat glioma cells were labeled with manganese oxide (MnO) and superparamagnetic iron oxide (SPIO) using electroporation and then visualized by MRI in vitro and in vivo in animal models. Labeling with MnO has been proposed as a good option for imaging the liver, melanomas, blood clots, hemorrhages and hemoglobin-derived regions with high concentration of endogenous iron to avoid the artifacts that may occur by SPIO-tagged cells [38] . Given the easy method of labeling, relatively low level of cytotoxicity, and efficiency in MR signal enhancement as the positive contrast agents, Mn-compounds may be a good alternative for cell tracking in the case of patients with kidney disorders.

2.3. Fluorine-Based MRI Contrast Agents

Fluorine is an element that is not found naturally in the human body. The ¹⁹F nucleus is therefore an extremely specific and also sensitive CA for MR imaging; its resonance is different from ¹H proton by just 6% which enables conducting ¹⁹F MR imaging on standard clinical MR scanners [39] . Perfluorocarbon (PFC) emulsions are nontoxic and biologically stable and can be produced for uptake by phagocytotic cells [40] or non-phagocytotic cells using various methods, including the use of TAs [41] . In fact, even non-phagocytic cells (T cells and neuronal stem cells) have been tagged with anionic or cationic ¹⁹F emulsions without the need for any additional TAs, which is important as most TAs are not approved for clinical use. Human neuronal stem cells (hNSCs) were labeled by simple incubation with perfluoropolyether (PFPE) nano-emulsion and successfully tracked after implementation into a mouse brain [42] . Human dendritic cells (DCs) labeled with a ¹⁹F agent showed no significant changes in viability, phenotype or functionality while used in vaccination tests [43] . Unfortunately, ¹⁹F MRI tracking in vivo involves quite a large number of cells [44] . However, a significant advantage of using ¹⁹F is that PFCs have been extensively studied as hemoglobin substitutes and the toxicity of these compounds is familiar [45] .

2.4. CEST/PARACEST MRI Contrast Agents

CEST (Chemical Exchange Saturation Transfer) agents exchange one or more ¹H protons with solvent water present in the body. PARACEST agents (Paramagnetic Chemical Exchange Saturation Transfer) are the novel group of ¹H-MRI CAs containing a paramagnetic lanthanide atom and allow simultaneous tracking of different cell populations in the same anatomical region by using a standard clinical MR scanner. Various types of functional groups with exchangeable protons have been used to improve PARACEST agents as tools for molecular imaging [46] [47] . In the recent study, murine macrophages and melanoma cells were labeled with highly biotolerable Yband Eu-HPDO3A by electroporation and subsequently were injected into healthy mice and tracked in vivo. This co-localized both types of cells so they could be seen in one image [48] .

2.5. Superparamagnetic Iron Oxide (SPIO) as MRI Contrast Agents

SPIO are the wide group of negative MRI contrast probes. They were first proposed to be MRI CAs for imaging the liver and spleen [49] -[51] but now they are probably the most extensively used MRI CAs; they provide labeled cells with a strong magnetic moment and have significant impact on T₂ and T₂*-weighted images. An iron atom with its 4 unpaired electrons, when placed in a strong magnetic field, aligns its magnetic moments to the direction of the magnetic field which generates a change in the MR signal. In contrast to paramagnetic metals, SPIO allows the visualization of a small number of cells.

There are various methods (e.g. coprecipitation, microemulsion methods, biomimetic mineralization etc.) used for preparation of SPIO, which result in a wide range of physicochemical properties [52] . Considering their size, iron oxides can be classified into three groups, based on their overall diameter: ultra small superparamagnetic iron oxides (USPIO, <50 nm), superparamagnetic iron oxides (SPIO, 50 - 150 nm) and micrometer-sized iron oxides (MPIO, >1 um) [53] . SPIO usually consist of an iron core and a hydrophilic coating. The core of an iron oxide nanoparticle consists of magnetite (γ-Fe₂O₃), maghemite (Fe₃O₄) or heamatite (α-Fe₂O₃) [52] (Figure 2). The first two types of nanoparticles (magnetite and maghemite) have been mostly used in biomedicine. The surface of the core must be coated in order to avoid the agglomeration of nanoparticles, improve their colloidal stability, prevent the oxidation, and make them more suitable for biomedical applications. There are several groups of compounds utilized for this purpose: polymers (e.g. dextran, carboxydextran, chitosan, PEG), organic surfactants (e.g. sodium oleate), inorganic metals, inorganic oxides (e.g. silica shells), and biological molecules (e.g. antibodies or liposomes) [52] .

Unlike gadolinium and fluorine, iron is an element naturally present in the human body and mammals are, in general, well adapted to maintain an iron balance. In Kupffer cells (macrophages/scavengers) and macrophages in spleenic red pulp, SPIO are quickly degraded and the released iron is incorporated into the normal blood iron pool (e.g. hemoglobin) [54] .

Phagocytotic cells, like macrophages, can usually take up and accumulate sufficient amount of SPIO and can be visualized by MRI [55] . Unlike the phagocytotic cells, most of the stem cells do not easily internalize iron oxide nanoparticles. There are several techniques for direct labeling of non-phagocytotic cells, which include simple incubation, transfection with polycationic transfection agents, coupling with antibodies, magnetofection, and magnetoelectroporation [53] [56] -[61] . Although SPIO are considered to have rather low cytotoxic level and not to have a negative impact on cell functionality [62] , there are suggestions that labeling with SPIO may lead

to alteration of cell function: in this study SPIO-labeled mesenchymal stem cells did not undergo the normal chondrogenesis pathway [63] . The number of studies making use of SPIO-labeling and investigating its efficiency and accuracy in monitoring the stem cells fate in vivo and in vitro is growing very fast. (U)SPIO have been applied for MRI monitoring in vivo among other things for: tumor lymphatic metastases [64] , human dental pulp stem cells (hDPSCs) transplanted into a mouse brain [65] , mesenchymal stem cells transplanted into a dog stroke model [66] , and mesenchymal stromal cells in a porcine heart [67] . The safety of SPIO-labeling has been proven in a clinical study on humans—human mononuclear cells were tagged in accordance with Good Manufacturing Practice (GMP) guidelines, injected into humans, and visualized by MRI [68] .

2.6. Reporter Genes as MRI Contrast Agents

Indirect tagging of cells utilizing genetic modification relies on incorporation into the cells of interest: 1) genes encoding receptors that specifically bind a CA, 2) genes encoding membrane translocation signals which enables the transport of a CA into the cytoplasm or 3) genes that encode enzymes chemically modifying the CAs. The main advantage of the reporter gene approach is stable detection of the cells as the CA does not dilute following cell divisions; on the contrary, dividing cells multiply, accumulate and thus provide an enhanced signal. Importantly, the gene expression is proof of cell viability, protein synthesis occurs only in living and active cells. The introduction of the selected reporter gene into a cell is possible using viral (e.g. lentiviruses) or non-viral methods (e.g. nanoparticles, polymerase chain reaction (PCR), commercially available cationic TAs or electroporation). Although in the case of viral methods integration of the gene into the genome of a cell does not require other TAs, which simplifies the procedure, there are serious health risks involved. For MR tracking, cells are programmed to express iron-containing proteins or other proteins that can be subsequently labeled with SPIO (e.g. biotin and streptavidin) [69] .

In the recent study [70] , human breast cancer stem cells (BCSCs) were transduced with dual reporter genes (human ferritin heavy chain (FTH) and green fluorescence protein (eGFP). Next, the transduced BCSCs were transplanted into mice. Importantly, no changes in the viability and functionality of the BCSCs were observed connected with ferritin over expression. MR imaging showed relevant differences in the signal intensities between normal BCSCs and FTH-BCSCs, both in vitro and in vivo. These results imply that ferritin-based MRI, which provides high spatial resolution and tissue contrast, has potential to be used as a technique to identify viable cell populations derived from BCSCs. Ferritin has also been used as an MR imaging reporter gene in another study, human mitochondrial ferritin (FTMT) was cloned and deprived of its mitochondrial targeting signal in order to accumulate in cytoplasm and load iron. Subsequently, the expression of the FTMT reporter gene was observed in vivo in mice [71] .

Also CEST-based reporter genes have been developed for MR imaging, e.g. human protamine 1 (hPRM1) and lysine-rich protein (LRP) [72] [73] .

3. Optical Imaging (OI)

Optical imaging (OI) methods include bioluminescence imaging (BLI) and fluorescence imaging (FLI). These methods are, like MRI, non-invasive, and enable long-term tracking of the cells, however, they lack MRI’s high spatial resolution. OI is based on two well-known phenomena: bioluminescence and fluorescence. The first term describes an active production and emission of light by living organisms and involves the oxidation of a specific substrate by an enzyme. Fluorescence on the other hand is the emission of light by a material that has absorbed electromagnetic radiation. BLI and near-infrared (NIR) FLI belong to very sensitive imaging modalities (10⁻¹² to 10⁻¹⁵ µm/L) but their limitation is the low level of tissue penetration of light, which corresponds to 700 - 900 nm and is called the “tissue transmission window” [74] [75] . Organic fluorophores, e.g. DAPI, rhodamin or fluorescein are inexpensive, simple to apply, and have tolerable cytotoxicity levels. However, the main disadvantage of these compounds is their susceptibility to photobleaching, pH fluctations, and chemical degradation [76] [77] .

The most popular reporter gene for OI is the encoding green fluorescent protein (GFP) and its derivatives, e.g. yellow fluorescent protein (YFP) and cyanian fluorescent protein (CFP). The photostability of GFP is better than in case of organic fluorophores. However, its low resolution and the risk of overlapping with the cell’s natural fluorescence are considered to be serious drawbacks in using these genes. Alternatively, red fluorescent protein, mKate, Katschuka protein or mCherry can be introduced into the cells [78] -[80] .

Quantum dots (QDs) may be a good alternative for organic fluorochromes due to their photostability, minimal bleaching, and possibility of adjusting their emission wave length. The main issue concerning QDs is quite high cytotoxicity profile, due to the presence of cadmium as a component and therefore their application in humans has been limited [76] .

Luciferases are a group of photoenzymes that emit photons during oxidation of substrats. To catalyze reactions, they require the presence of ATP and oxygen. The best aspects of tracking the cells with this method are the stability of transfection and the long-term monitoring of cell viability [81] . The major problem with optical imaging is the abovementioned low tissue penetration of light; this limits OI alone to preclinical investigations in cell cultures or small animals.

4. Nuclear Medicine

Particular attention has been given recently to the tomographic methods: PET and SPECT as cellular imaging modalities. For the application of these methods, tagging of cells with radiotracers is required. The high efficiency of PET and SPECT allows using very small amounts of radioactive CAs, which are considered safe for use in humans. The main advantage of SPECT is the possibility of simultaneous dual-tracer imaging of isotopes with different energy levels, i.e. monitoring two different isotopes at the same time, which enables distinguishing between two populations of cells. In the case of PET, dual-tracer imaging is difficult because all PET-radiotracers have the same energy level. However, the most important advantages of PET over SPECT imaging include higher sensitivity (two or three orders of magnitude) and FDA approval as a clinical imaging technique in humans [82] -[84] . Radionuclides with a relatively long half-life time are applied in direct labeling of cells for nuclear medicine, such as ¹¹¹In, ⁶⁴Cu, ¹⁵O, ¹³N, ⁶⁸Ga and ¹⁸F. The radionuclides are coupled with other compounds, e.g. glucose [85] and incubated with the cells of interest. The great majority of studies did not show radiotracers containing ¹¹¹In or ⁶⁴Cu to have any serious undesirable effects on function and differentiation of cells. However, there is some suggestion that ¹¹¹In-containing tracers can alter the function of hMSC [86] .

As mentioned above, in most cases direct cell labeling does not give information about the cell viability and one approach to bypass this problem is indirect labeling by introducing a reporter gene encoding a tracer. One of the most commonly used reporter genes for PET are wild-type herpes simplex virus type 1 thymidine kinase (HSV1-tk) and its HSV1-sr39tk mutant that have been applied for monitoring tumor-specific lymphocytes in mice. By phosphorylation of the radionucleoside analogs and adding a negative charge on their surface, thymidine kinases allow the cytoplasmic agglomeration of the CA. However, these compounds are not able to pass the blood-brain barrier (BBB) which disqualifies them as intracerebral cells tracers [87] [88] . A human deoxycytidine kinase containing three amino acid substitutions within the active site (hdCK3mut) was utilized as a reporter gene in combination with the PET probe [18F]-L-FMAU to monitor mice models with human hematopoietic stem cell (hHSC) transplantation. Long-term measurements of the engrafted cells (up to 32 weeks) demonstrated that hdCK3mut expression is stable in vivo [88] . Other reporter genes proposed for PET and SPECT imaging include: the thyroidal sodium iodidesymporter (NIS), 376aa, somatostatine receptor type 2, dopamine D2 receptor [89] [90] , and hERL/18F-FES [91] .

5. Photoacoustic Imaging and Ultrasound

Photoacoustic imaging (PAI) is a non-invasive and nonionizing imaging method relying on the photoacoustic effect. Biological samples are exposed to short laser pulses and absorb the light differently due to the differences in their chemical composition. Next, the absorbed energy is converted into heat, which generates an ultrasonic acoustic signal [92] . Two types of photoacoustic imaging have been developed: PAM (photoacoustic microscopy) and PAT (photoacoustic computed tomography) [93] . Both, endogenous and exogenous labeling is used for PAI. Blood vessels and tumors can be imaged using in vivo labeling. Additional tracers, such as methylene blue have been used for imaging of lymph nodes. Gold-based nanostructures-nanoparticles or nanorodes are considered and tested as PAI contrast agents because of their low toxicity profile as well as sizeand shape-dependent features. Also, carbon nanomaterials, e.g. modified single-walled carbon nanotubes (SWCNTs) are applied in photoacoustic imaging due to their optical properties [94] .

Ultrasound (US) is a well established clinical imaging procedure. Contrast in the US imaging is generated by sound waves penetrating tissues which differ in density. Tracers for US imaging include microbubble-based or emulsion-based CAs (<10 µm in diameter) [95] which are composed of proteins and polymer shells containing

gases, e.g. air, nitrogen or perfluorocarbons. Also, silica, gold or polystyrene nanoparticles serve as US contrast probes [94] . As one of the recent studies shows, when Au-NTs-labeled mesenchymal stem cells (MSCs) are transplanted into tissue, photoacoustic imaging can detect the presence of the transplanted cells with sufficient penetration depth (10 - 50 mm) and spatial resolution (20 - 300 mm). Furthermore, it was possible to monitor Au-NTs labeled MSCs for an extended period of time (up to 2 weeks) with high sensitivity (102 - 103 cells) [92] .

6. Conclusion

Long-term monitoring of cells requires a reliable and efficient method of labeling and a sensitive method to enable their detection. Each of the described imaging modalities exhibits unique advantages and drawbacks. Table 1 represents the summary of in vivo imaging modalities and contrast agents. MRI and PET are the modalities which will probably be most extensively used for cell tracking due to their high spatial resolution and potential for clinical translation.

Acknowledgements

The research leading to the results in this paper receives funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement no 242175 and from the Sonnenfeld Stiftung, Berlin, Germany.

References

NOTES

^*Corresponding author.