Since DWI has been proven to be a critical functional technology in detecting and locating optic nerve disease lesions, for example, the use of DWI and the calculation of ADC values for evaluating optic neuritis (ON) has been reported [24, 25]. Using only DWI has high sensitivity and specificity in distinguishing acute from chronic ON. Differences in ADC values can reflect different pathogenesis of ON , and ADC values give a measure of axonal disruption in the chronic optic nerve lesion [27, 28]. Moreover, DWI has been used for the assessment of various diseases, such as trauma  and ischemia [30, 31], that involve the optic nerves. DWI can also be considered the first technique capable of identifying posterior ischemic optic neuropathy (PION) by identifying acute ischemic lesions of the optic nerve . And ADC may serve as a useful tool for prognostication for Optic pathway glioma (OPG), a significantly higher mean ADC was seen in OPG that required therapy for tumor progression .
But DWI of the optic nerve is difficult in clinical circumstances, on account of the small dimension of the nerves, uncontrolled eye movements, and the high signal from cerebrospinal fluid or neighbouring fat within the orbital region . ss-EPI DWI suffers from magnetic susceptibility artifacts, chemical shift artifacts, and low image quality. The present study found that rFOV DWI exhibited superior performance compared with ss-EPI DWI in all evaluated aspects, including blurring effects, image distortion, artifacts, lesion conspicuity, and image quality. Barker et al. also found that rFOV DWI provided improved subjective image quality of optic neuritis compared with ss-EPI DWI . Owing to the readout-segmented k-space acquisition strategy, rs-EPI effectively reduces the image distortion caused by the large susceptibility variations and the T2* blur effect . Therefore, the application of ss-EPI in optic nerve imaging has been limited in the past. In this study, we evaluated new technical approaches of DWI of the optic nerve.
rs-EPI DWI showed higher reproducibility than did ss-EPI DWI and rFOV DWI. In the present study, rs-EPI DWI appeared to be the most robust and reliable method. The rs-EPI arrangement partitions the k-space trajectory into numerous portions in the readout direction. Accordingly, TE and encoding times can be decreased, and movement correction can be performed utilizing a 2D navigator correcting motion-induced, non-linear phase errors [36, 37]. However, the problem is that the acquisition time is longer on account of multiple TR intervals. Seeger et al. found that rFOV-EPI (2:45 s) showed improved image quality, the most accurate tumor delineation, and the best differentiation from retinal detachments compared with ss-EPI (2:14 s) and rs-EPI (3:07 s) in patients with uveal melanomas . In the present study, the reproducibility of the mean ADC value of rFOV DWI (2:03 s) was better than that of ss-EPI DWI (2:11 s), but the reproducibility of ROI2 was lower than for rs-EPI DWI (3:12 s). The mean ADC value of rFOV DWI was significantly lower than those of rs-EPI DWI and ss-EPI DWI, which is consistent with the results of Seeger et al. . We speculate the reason for these findings may be that we applied FOV rotation to remove the potential folding artifacts of rFOV-DWI and used complex averages to improve the ADC estimation. Furth more, rFOV DWI overcomes the major problem of low specific absorption rate and low spatial resolution facing DWI on the optic nerve as the reproducibility of the ADC values of the ss-EPI DWI sequence is relatively low. It is not recommended to use the ADC values of the ss-EPI DWI sequence as an indicator in follow-up cases.
Common diseases of the optic nerve include optic neuritis, ischemic optic neuropathy, optic nerve tumors, etc. The clinical manifestations and involvement of optic nerve segments in different diseases are different. For example, inflammation of the optic nerve can be involved unilaterally or bilaterally, and it can also be involved in long segments. In the application of the ADC values of the optic nerve, the ADC value of the contralateral side of the diseased optic nerve is usually used as the control region. Still, the ADC value of the area with no obvious abnormal signal is usually measured in time for bilateral involvement as a reference. ADC measurement of the optic nerve is challenging due to the small diameter of the optic nerve because there is the potential of partial volume averaging with surrounding CSF, fat, and osseous structures. Susceptibility artifacts caused by the air-bone-tissue interface also affected the ROI placement over the optic nerve. Therefore, three ROIs in different positions were placed to explore the possible differences in ADC values. The comparison of ADC values shows that the ADC value of ROI1 has the best reproducibility and is significantly higher than the ADC values of ROI2 and ROI3, indicating that the signal of the optic nerve behind the global is continuously affected by partial volume averaging with the high signal of the eyeball and needs to be taken into consideration to rule out false positives. And when choosing the control region for comparison, the area close to the eyeball should be avoided. There are both good reproducibility in ADC values of ROI2 and ROI3, and the difference between ADC of ROI2 and ROI3 is not statistically significant, which may indicate that compared to ROI2, ROI3 does not suffer more inaccurate due to susceptibility artifacts caused by the air-bone-tissue interface and partial volume averaging with the surrounding air and osseous structures.
Distortion creates a challenge for diffusion MRI of the optic nerve, especially at high field strengths. The optic nerve is surrounded by fat, muscle, bone, and air leading to large susceptibility changes that distort diffusion-weighted EPI images. Our research found that rs-EPI DWI showed the highest agreement with T2w imaging in terms of the length and angle of the optic nerve compared with those obtained with rFOV DWI and ss-EPI. The distortion in the length of the optic nerve may indicate the degree of compression of the image in the long axis of the optic nerve. The distortion of the angle of the optic nerve may represent the distortion ratio of the x-axis compared with the y-axis. Previous studies have reported quantitative evaluations of the degree of distortion in this process. In the study of Thierfelder et al. , the rFOV DWI of the prostate showed a stronger correlation with the T2w images in the coronal and sagittal diameters as well as in the prostate volume, compared with those obtained by ss-EPI DWI, yielding ICCs of 0.948 for the coronal diameter, 0.858 for the sagittal diameter, and 0.938 for the prostate volume. These observations are consistent (to a certain extent) with the results of the present study. Specifically, we observed that rFOV DWI has a good correlation with T2w images in terms of angle. Still, there is distortion in the long axis of the optic nerve, which is like the fact that the ICC coefficient of the prostate also is low in the longitudinal axis direction. We speculate that the lower ICC coefficient in the longitudinal axis may be due to the susceptibility artifacts caused by the air-bone-tissue interface.
There are several limitations in this study. Firstly, our quantitative analysis was restricted to a global comparison of the optic nerve length and angle. It might be of interest to determine how well other parameters in rFOV DWI or rs-EPI DWI correlate with T2w imaging. Additionally, the present investigation only evaluated the intraorbital segment of the optic nerve and did not examine other segments of this nerve. We have made the comparison between two product DWI sequences and a prototype ZOOMit DWI sequence; there are still some other techniques to reduce distortion and improve image quality that we haven’t compared, like BLADE DWI and ss-EPI with two b0 acquisitions of opposite phase-encoding directions, etc. Secondly, the methods tested in this study are not yet widely used in clinical practice. So, it seems that they may not be available on the specific vendor’s platform. Furthermore, the performance results of the three sequences and the results of the image quality assessment may vary depending on the specific hardware. Therefore, the findings reported in this study are specific to the hardware of these tests. Thirdly, the number of patients examined was only adequate for an exploratory study; additional work with a larger population and patients with pathology is warranted to corroborate our findings. Although ADC values of different methods are compared in volunteers, there is no reference to the gold standard, so ADC phantom can be further used to verify the standard in future experiments.