MRI of the Human Eye Using Magnetization Transfer Contrast Enhancement

Abstract

purpose. To determine the feasibility of using magnetization transfer contrast–enhanced magnetic resonance imaging (MRI) to track cataractous lens changes.

methods. A fast spin–echo sequence was modified to include a magnetization transfer contrast (MTC) preparation pulse train. This consisted of twenty 8.5-msec sinc pulses, 1200 Hz upfield from the water resonance and 1.2-Hz power. The MTC preparation pulse was followed by acquisition through fast spin–echo imaging. The imaging parameters were number of excitations (NEX) = 1, echo time (TE) = 14 msec, recovery time (TR) = 2 sec, echo train length of eight echos, and a matrix size of 256 × 160. To reduce motion artifacts, the volunteers were asked to fixate on a blinking LED. Normal and MTC-enhanced images were acquired from normal volunteers and volunteers with nuclear or cortical cataracts.

results. The eye was adequately imaged, with few motion artifacts appearing. The lens was well resolved, despite the short T₂. The cornea and ciliary body were also clearly visible. In the lens, resolution of the epithelium and cortex were enhanced with MTC. In addition, contrast-to-noise ratios were measured for each image. Examination of the contrast-to-noise ratio confirmed that MTC increased the contrast between the nucleus and cortex. Unenhanced MRIs showed significant differences between the cortex of normal volunteers and volunteers with cataracts. MTC-enhanced images improved the sensitivity to changes in the nucleus.

conclusions. In this preliminary study, we were able to use MTC-enhanced MRI to obtain high-contrast images of the human lens. Regular and enhanced MRIs detected statistically significant differences between normal and cataractous lenses.

Magnetization transfer contrast (MTC) enhancement is a magnetic resonance imaging (MRI) technique that takes advantage of the magnetic interactions between water and macromolecule hydrogen atoms. ¹ The signal used for MRI primarily originates from free (bulk) water protons. Underlying this signal is another originating from hydrogen atoms in the tissue macromolecules. This signal is not easily observed directly because of the large bandwidth of the macromolecules, relative to the free water protons. However, the macromolecular pool can be selectively saturated by applying an off-resonance radio frequency irradiation. If the water protons and macromolecular protons are capable of magnetization exchange through dipolar and/or chemical exchange, then the applied saturation can be transferred to the bulk water protons. This results in a signal decrease in the MR image where an effective coupling between the bulk water and the tissue macromolecules exists. It has previously been shown that this effect is dependent on the concentration, mobility, and surface chemistry of the macromolecules. ² Thus, MTC is sensitive to edema that alters the concentration and mobility of macromolecules and remodels macromolecular structures within the tissues. 

Lens opacifications must be accompanied by alterations in tissue compositions including the hydration state. MTC-enhanced MRI has been successfully used to document lens changes in longitudinal studies of galactosemic dogs during sugar cataract formation. ^{3 4 5} These MTC-weighted images were consistent with localized osmotic lens changes during cataract formation and consistent with the osmotic hypothesis of sugar cataract formation. ⁶ Images with MTC enhancement were able to show lens changes sooner than with standard MRI sequences. Moreover, region of interest (ROI) analysis of the images revealed that significant tissue changes occurred before any clinically visible lens changes. 

Although a number of studies have reported using conventional MRI in the eye, ^{7 8} the purpose of the present study was to investigate the feasibility of developing high-resolution techniques for collecting MTC-weighted MR images of the human eye. Because the eye is a freely moving organ located within a bony structure that limits access to surface coil and its resultant magnetic signal, a number of problems had to be explored in this study. These included obtaining adequate high-resolution images of the fine physiological structures of the eye with minimal motion artifacts while imaging in the presence of high-susceptibility gradients around the eye with adequate signal-to-noise levels. To overcome some of these issues, a custom surface coil was developed to adapt to the eye anatomy along with a visual fixation scheme to minimize eye motion. A segmented fast spin–echo imaging method that permitted rapid high-resolution imaging of the eye was used to overcome the limitations of the magnetic field variations around the eye. 

Materials and Methods

This research adhered to the Declaration of Helsinki for research involving human subjects. All normal volunteers and patients with cataract participated in an National Eye Institute Institutional Review Board–approved clinical protocol and gave informed consent. They all also underwent a complete comprehensive eye examination including slit lamp examination and grading of their lenses using the Lens Opacities Classification System (LOCS) II. ⁷ Pupils were dilated maximally for the eye examination but not for the MRI examination. All imaging experiments were performed on a Signa 1.5-T (General Electric, Milwaukee, WI) MRI system. Images were acquired using a custom-made 4-cm receive-only surface coil attached to a specially constructed Lucite framework with a locking headband. This allowed the probe to be securely and comfortably locked to the volunteer being scanned. The headband was also equipped with a blinking LED (∼1 Hz) fixation target positioned approximately 6 in. above the eye being scanned. Spin excitation was accomplished through the body coil. 

MR images were acquired from the right eye only. There were five normal volunteers, all of whom had LOCS II scores of nuclear color (NC) = 0, nuclear opalescence (NO) = 0, cortical opacity (C) = 0, and posterior subcapsular cataract (PSC) = 0. There were four pure nuclear cataracts. Two of these had an LOCS II score of NC = 2, NO = 2, C = 0, and PSC = 0, and two had an LOCS II score of NC = 1, NO = 1, C = 0, and PSC = 0. Last, there were five pure cortical cataracts, four with an LOCS II score of NC = 0, NO = 0, C = 2, and PSC = 0 and one with an LOCS II score of NC = 0, NO = 0, C = 1, and PSC = 0. The average age of the normal volunteers was 63 years, and the average age of the patients with cataract was 58 years. Images were acquired using a fast spin–echo sequence modified to provide either an MTC preparation pulse or a dummy pulse of the same length but with no radio frequency (RF) power. The MTC preparation pulse consisted of twenty 8.5-msec sinc pulses, 1200 Hz upfield from the water resonance and 1.2-Hz RF power. The image parameters were 14-msec echo time (TE), 2-sec recovery time (TR), 1.5-mm slice thickness, 80-mm field of view (FOV), 256 × 160 matrix size, and an echo train length of eight echos. Real-time phase correction was performed by adjusting the readout dephaser. 

An initial axial locator scan was performed with a multislice, fast gradient recalled echo sequence. The lens center was located and used to prescribe a set of sagittal slices for the MTC fast spin–echo sequence. The same set was acquired several times with (M_s) and without (M₀) the MTC preparation pulse. The set of images least affected by motion artifacts was used for ROI analysis. 

Contrast-to-noise ratios were measured for several ROIs in the lens. The contrast-to-noise ratio is defined by  

\[C{=}\ \frac{I_{\mathrm{ROI}}-I_{\mathrm{Ref}}}{N}\]

where I _ROI is the average intensity in the ROI, I _Ref is the average intensity in a reference region (either vitreous or aqueous), and N is the root mean square (RMS) noise in the image. The contrast-to-noise ratio provides a convenient, meaningful, way of comparing tissues within and between images. The aqueous was used as the reference region, because its signal should not be altered by either cataract formation or MTC contrast enhancement. 

Statistical significance was determined using the Students’ t-test for unpaired samples with unequal variances. 

Results

For this preliminary study, images were acquired from volunteers between the ages of 49 and 78 years. Before MRI, each volunteer was examined by conventional ophthalmologic methods to ensure that the eyes were normal, and the lenses contained no opacities. Each MR imaging examination lasted approximately 45 to 60 minutes. During each session, an initial set of images was acquired using a fast, multislice, gradient recalled-echo sequence. These images were used to determine a graphic prescription for a set of sagittal slices through the center of the lens. Each set was collected several times with (M_s) and without (M₀) MTC enhancement. A typical examination had four repeated images for M₀ and M_s but could comprise as few as three or as many as seven. 

Figure 1 illustrates typical MR images obtained from a volunteer. The M₀ images give detailed anatomic pictures of the eye including the lens. The lens, cornea, ciliary body, and miscellaneous muscles were readily distinguishable. In addition, a ring of higher than average intensity appears in the lens, separating the adult nucleus from the lens cortex. This demarcation between the lens cortex and nucleus was improved when MTC enhancement was applied. Although the total signal from the lens is decreased, it is still nonzero. The slight blurring observed in both images is due to motion effects, such as blinking and vision wandering away from the fixation target. 

ROI analysis was applied to several parts of the lens using custom software (written for the Interactive Data Language software package; Research Systems, Boulder, CO). When the images with the fewest motion artifacts were used for analysis, the average variation between image intensities in selected regions as described below was less than 10%. The ROIs in the lens were defined using a mouse-driven program. These consisted of the entire aqueous, which was used as a reference region, the anterior cortex and epithelial layer, the posterior cortex, and lens nucleus. The cortical region was identified by tracing the lens perimeter and the outer edge of the ring demarcating the adult nucleus. The region was determined to be either anterior or posterior, on the basis of its location relative to the geometric center of the lens perimeter (e.g., anything forward of the center is anterior). The lens perimeter was scaled down to 25%, and the area inside was used for the nuclear ROI. The ROIs were used to calculate contrast-to-noise ratios (according to the equation shown in the Methods section) that were used to determine the extent of MTC enhancement. 

Results from ROI analyses of the MRIs from each group were tabulated and statistically compared. As a test of reproducibility, images acquired from the same volunteer were com-pared with each other. No significant statistical difference between their ROI values were observed. However, statistically significant differences in the ROI data from unenhanced MR images of normal and cataractous lenses, summarized in Figure 2 , could be observed. Taking P < 0.10 as significant, nuclear cataracts showed significant changes (P < 0.09) in the posterior cortex of the lens, and cortical cataracts showed significant changes (P < 0.08) in the anterior cortex of the lens. In contrast, the ROI data from enhanced normal and cataractous lenses, summarized in Figure 3 , showed no significant differences in the cortical area but increased difference in the nucleus. In the MTC-enhanced MR images, the P for the Student’s t-test for the nuclear region of lenses with nuclear cataracts changed from 0.30 to 0.17. 

Discussion

High-resolution images of the eye can be obtained with standard and MTC-enhanced fast spin–echo techniques, despite the effects of blinking and other involuntary eye motion. Compared with standard T₂-weighted images, contrast in the present MR images was reduced by the short TE required to visualize the lens. Nevertheless, the ciliary body, cornea, and miscellaneous muscle tissue were identifiable in both types of images. 

Image fluctuations resulting from eye movements were minimized through the collection of multiple slices using continuous pulsing of the MRI gradients. Images collected in this way generally had fewer motion artifacts due to flinching. However, a major disadvantage of multislice imaging is the introduction of additional MTC effects in adjacent slices from the slice selection pulses. ⁸ This does not appear to be a limiting factor, as long as the slice selection parameters remain fixed. In principle, these slices can be used for three-dimensional reconstruction of the lens. 

ROI analysis of the images indicated that the MTC images demonstrated a larger amount of contrast between the lens and the aqueous. These data are consistent with previous in vivo studies in dogs. ⁴ Contrast-to-noise ratios in the cortex were similar for humans and beagles, but the value for the nucleus in humans was one sixteenth that measured in beagles. ⁴ This could be due to species differences, aging, or the more conservative irradiation power used in the human studies. 

The contrast-to-noise ratios measured for the anterior cortex were always slightly higher than those measured for the posterior cortex. This is due to a combination of two factors. First, the anterior measurement includes the epithelial layer. The tissue properties in that layer are different and contribute to a shift in intensities. Second, the receiving range of a surface coil drops off with distance. There is a slight decrease in signal but no decrease in noise. Because the contrast is signal dependent, the contrast-to-noise ratios decrease. 

Cataractous changes in the lens can be observed in both normal and MTC-enhanced MR images. Cortical ROI changes were better observed in unenhanced images, whereas increased sensitivity to nuclear changes was observed with MTC enhancement. The addition of an MTC preparation pulse suppressed contrast changes in the lens cortex. Compared with the more protein-dense nucleus, contrast sensitivity in the cortex primarily comes from bulk hydrogens on the free water molecules. This results in a reduced sensitivity to MTC, because the available surface area is decreased and saturation transfer is reduced. For the more protein-dense nucleus, the MTC can more readily be observed, and this can result in increased sensitivity in observing cataractous changes. In the nuclear cataracts, the observed changes in MTC signal were probably due to altered protein hydration associated with protein cross-linking. Whereas P for the nucleus in nuclear cataracts was improved by MTC, differences were not significant. This may be due to an incomplete saturation. Further studies are needed to determine the proper MTC presaturation, and such studies should control for the effects of aging. 

Conclusions

The addition of an MTC preparation pulse to a standard MRI sequence yields high-contrast, detailed pictures of the lens. These preliminary studies suggest that cortical changes can be better observed with unenhanced MR images, whereas nucleus changes are better observed by addition of the MTC preparation pulse. 

 Submitted for publication February 3, 2000; revised May 16, 2000; accepted June 2, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

MRI of the human eye without (A) and with (B) MTC enhancement.

Figure 2.

 

Summary of ROI data from unenhanced MR images of normal and cataractous lenses. P was determined using a Student’s t-test. Columns represent average value for each group. (•) Anterior cortex; (▴) posterior cortex; (▪) nucleus; NS, not significant.

Figure 3.

 

Summary of ROI data from MTC-enhanced MR images of normal and cataractous lenses. P was determined using a Student’s t-test. Columns represent the average value for each group. (•) Anterior cortex; (▴) posterior cortex; (▪) nucleus; NS, not significant.