Magnetic resonance imaging (MRI) is a medical imaging device widely used for investigating anatomical and physiological information from in-vivo. The important role of MRI is acquiring images with non-invasive way. Over the last decade MR imaging at ultra-high fields (UHF) is becoming increasingly common. Stronger magnetic field leads to a higher signal-to-noise ratio (SNR) and allow for high spatial resolution imaging. With 7T, we can acquire sub-millimeter structural image in a clinically reasonable scan time^[1]. Moreover, if longer scan time is allowed (e.g. ex-vivo scan) super high resolution (scale of 200μm) imaging is achievable. And it could provide comparable image resolution to histological sections. In general, relaxation time constants (T₁ and T₂) of newly fixed tissue change over time, particularly in the early stage of tissue fixation^[2]. High-resolution imaging of small tissue samples in large bore MRI needs longer scan times because signal averaging is necessary. Therefore, image contrast can change over time. In this work, we have 1) performed MRI scans using human globe specimens to investigate the contribution of fixation to the evolution of T₂ over time and 2) measured SNR changes as function of the number of excitations(NEX) to gain some insight into the optimal average number. We also show a high-resolution ex-vivo globe image using 7T.Shortly after enucleation of the eye-globe (within 3 hours) from a patient who has choroidal melanoma on the left eye, the specimen was transferred for MR imaging. The globe was fixed in 10% buffered formalin. Data were collected using 7T MRI (Siemens; IRB approved). To examine the influence of fixation on T₂ of the tissue, 2D single-slice multi-echo spin-echo data were acquired with parameters as follows: TR = 1500ms, TE = 9.7 to 145.5ms (echo spacing = 9.7ms), number of echoes = 15, in-plane resolution = 0.5×0.5mm², slice thickness = 2 mm and 25 different time point data were collected over 23 hours. The averaged T₂ values and its standard deviation within the ROI were calculated. Slice thickness of the refocusing gradient has been increased to reduce stimulated-echo signal from multiple-refocusing^[3]. When the evolution of T2 showed limited change, structural imaging of the globe was performed. MP2RAGE^[4] sequence (Siemens) was used with following parameters: TR/TE = 6000/4.42ms, TI1/TI2 = 700/2700ms, resolution = (250um)³-isotropic voxel, scan time = 21.02 min and 25 averages. The first inversion time was optimized to enhance tissue boundary. The ΔSNR(n) (=[SNR(n)-SNR(1)]/SNR(1)) and its standard deviation within the ROI were measured.

Figure 1 shows the T₂ changes affected by fixation. Four ROIs, shown in Fig. 1A, were used for the analysis. When the T₂ spectra were plotted over time, each plot demonstrated different amount of variation. ROI1 shows significant T₂ reduction while increasing in ROI3. Any noticeable changes were not seen in ROI2 and ROI4. Twelve hours later, all the ROIs showed limited T₂ changes.

ROIs in the specimen and ΔSNR changes in the 1st and 2nd inversion MP2RAGE data over different number of NEX are plotted (Fig. 2A). The signal plot shows large SNR changes in the first 10 averaged results (NEX ≤ 10) while it is limited (less than 3% differences) when the NEX is larger than 10. When the averaged images were qualitatively investigated, there was not a significant SNR changes from NEX = 11. These results suggest that the minimum number of image averages at current resolution is 11.

Figure 3 show high-resolution ex-vivo globe images from 7T MRI. Structures of eye-globe such as lens, vitreous body, cornea, iris, anterior chamber, retina, outer sheath of optic nerve, optic nerve and central retinal vein were clearly identified. Moreover, choroidal melanoma was also successfully detected (red arrows) and it is well corresponded with histopathology result (Fig. 4).

In this work, high-resolution ex-vivo globe scan was performed using 7T MRI. With voxel size of 250μm isotropic resolution, detailed structures of the globe were clearly identified. Our results show that the high-resolution MR imaging is particularly useful in helping to visualize inner structure of eye-globe with non-invasive way. Compared result with histopathology suggests a high degree of similarity. Moreover, it is successfully demonstrated the sensitivity to choroidal melanoma. The results from our study suggest that eye-globe shows limited T₂ signal changes 12 hours after the tissue fixation. In addition, with the imaging parameters used in this study the adequate number of image averages to acquire comparable SNR is larger than 11. In conclusion, this study has shown that the super-resolution MR images provide a useful contrast to delineate the ex-vivo tissue structures with non-invasive way.