Functional magnetic resonance imaging (fMRI) is a well-established technique in visual neuroscience, but is not widely used in ophthalmology or neuro-ophthalmology practice, despite the information it can provide about neural function and dysfunction in patients. Here, we review some of the ways that we have used fMRI to characterise visual dysfunction, and discuss fMRI generally as a candidate for assessing visual function clinically.

The studies described largely take advantage of the fact that visual cortex is arranged retinotopically as Holmes revealed almost a century ago [1]. Selectively stimulating different regions of the visual field allows multiple visual field mappings to be identified reliably in individuals with fMRI (as shown in Figure 1, see also refs [2,3]). This ability to produce informative data from a single individual is vital for a clinically viable tool and sets this method apart from many other fMRI methods. We have used the retinotopic mapping procedures to characterise visual dysfunction that arises from a number of different causes, at different stages in the human visual system.

Figure 1: Retinotopic maps in a normally sighted individual. Top: the two stimuli that systematically move through eccentricity (left) and polar angle (right). Such stimuli elicit a travelling wave of activity in cortical retinotopic representations. The location in the visual field that elicits activity is colour coded (as indicated in the key to the right
of the stimuli). When this colour coding of activity is presented on 3-D (middle row) and flattened (bottom row) surface reconstructions of the cortex, multiple visual maps in the occipital lobe can be identified. To the left and right data are given for activity elicited by rings and wedges, revealing maps of eccentricity and polar angle, respectively. On the flattened reconstructions the early visual areas, V1, V2 and V3, are outlined, demarcated by the representations of the vertical and horizontal meridians, visible on the polar angle maps (right).

We have used the retinotopic mapping procedures to characterise visual dysfunction that arises from a number of different causes, at different stages in the human visual system.

Retinal lesions

Recently, we assessed cortical signals in patients with retinal lesions. We were motivated by neuroscientific questions concerning reorganisation of cortical maps, something previously reported in patients who are born without functioning foveal photoreceptors [4]. We tested a relatively large number of patients with macular degeneration (those with Stargardt’s disease and the more common age-related form), finding that the cortex does not remap visual information in these patients [5]. This allows for the intriguing possibility that assessing cortical signals might helpfully determine retinal sensitivity in ophthalmological disease; the activity at a particular spatial location within the map should predict visual sensitivity. This might be used in a number of ways. We are currently assessing the effectiveness of antiangiogenic treatment for age-related macular degeneration, and changes in activity in the calcarine map largely reflect the changes in conventional visual measures of acuity and perimetry [6]. One of the advantages of measuring cortical signals is that the map is fixed, so no matter where the eyes are pointing, if the retina is stimulated (with a Ganzfeld stimulus for instance) activity in the cortex must reflect intact retinal processing. This is helpful in patients who find it hard to fixate for perimetry measurements.

Cortical lesions

We derived visual maps in the early visual areas of a patient, GY, who exhibits residual visual capacities in his right, “blind” visual field following a lesion to the calcarine cortex in the left hemisphere [7]. This patient displayed maps in early visual areas that largely reflected his scotoma, but subtle differences in the mapping of extrastriate regions were also evident when only the regions of the scotoma were stimulated. This modified mapping in the early visual cortex could be explained on the basis of changes in local connections, indicative of reorganisation. Reorganisation in this case may not come as a surprise given that the patient was tested over 30 years after a lesion sustained at the age of 8. Interestingly, recent evidence from diffusion tensor imaging also seems to indicate reorganisation of the visual pathways in this patient [8].

Over recent years we have been consulted by neurosurgeons, keen to evaluate the potential visual effects of removing lesions close to the visual representations of the brain. In one such case a mass was located in a lateral area of the occipital pole. Although the location of visual maps is broadly consistent across individuals, the specific location of the representation of the central visual field varies considerably. Our mapping procedures showed that the mass was proximate to representations of the fovea in early visual areas allowing the surgeon to gauge the potential impact of the procedure on visual function. In addition to the primary visual cortex more than twenty extrastriate visual maps can be identified. Many of these have specific roles in visual perception, with localised lesions giving rise to specific visual deficits such as cerebral achromatopsia [9,10]. Knowing the spatial relationship between a potential site for surgery and a patient’s extrastriate visual maps allows the impact of surgery on perception to be predicted.

Afferent visual pathways

Some developmental disorders can give rise to abnormal routing of visual information from the eye to the brain, which would escape detection with anatomical imaging. In individuals with albinism the decussation of the optic nerve at the optic chiasm is abnormal with a larger number of fibres originating from the temporal retina crossing to the hemisphere contralateral to the eye. This gives rise to abnormal lateralis`ation and mapping of signals in the visual cortex. We found that in humans the cortical map of the abnormal retinal input coexists with the map of the normal retinal input such that a region of primary visual cortex will respond to equal but opposite locations in the ipsilateral and contralateral visual fields.11 When we evaluated how fMRI might be used to detect the presence of albinism [12,13], we found that fMRI is very effective for detecting the abnormal lateralisation of visual signals. Indeed it was equal to if not better than current electrophysiological methods used clinically.13 Detecting albinism can be a challenge because pigmentation alone is not diagnostic.

When lesions occur in white matter, visual field defects are less straightforward to predict on the basis of anatomical imaging alone. We described how the brain signals we measured in response to visual stimuli could be used to generate a map of visual locations to which the brain responded in one such patient [14]. In other words, we could use the brain activity to predict where the patient could see. The individual in question had undergone surgery to remove a left hemisphere parieto-occipital mass. The medial occipital cortex was intact following surgery, but the patient complained of difficulty reading and initial perimetry measurements pointed only to a questionable central field defect. Our fMRI mapping experiments indicated that left calcarine cortex no longer responded to a central (<3 deg) region of the lower right quadrant. Following up on this finding, finer perimetry measures of central visual field revealed a scotoma in the predicted location. It was likely therefore that the surgery resulted in a lesion of the optic radiation.

| Sign up to the ACNR email newsletter here |

In more recent work we have looked at a series of patients with known optic radiation lesions due to stroke. We followed one case over time and here we report on the changes in the patient’s visual sensitivity and cortical activity. Figure 2 shows that the visual field recovers, particularly within the red circled region. The visual field maps represented on the surface reconstructions of the occipital lobe also exhibit a change between the examinations. Specifically, the part of dorsal primary visual cortex (V1) that would normally represent the visual field defect exhibits a disrupted map that later becomes largely normal, reflecting the change in visual sensitivity. As with the previously discussed case, the cortex initially fails to respond strongly to the region of the field deficit, but in this case as visual sensitivity recovers, so too do the cortical signals. It is noteworthy that the cortical representation of the field deficit is not silenced, but instead responds to a different field location. This could be interpreted as reorganisation, but the alternative explanation, which we currently favour, is that the normal receptive field properties of neurons could lead to the signals we record (see refs [4,5]).

Figure 2: Visual fields as assessed behaviourally, left, and with fMRI, right. Data are given for two exams (first – top; second – bottom) separated by 12 weeks. At first the scotoma is dense in the lower left quadrant, but becomes less so by the second exam. This change is mirrored by a
change in the cortical map in primary visual cortex, V1 (arrows). The map originally represents central locations but not peripheral (coded blue). Later the more peripheral locations are mapped.

What issues have been and need to be overcome to allow fMRI to be translated into clinical assessments of vision? First, reliable information must be obtainable from a single individual. This has been largely overcome, however it still remains to document carefully the reliability of signals within an individual from session to session and within sessions. Second, and related to reliability, the time it takes to gather sufficient data to characterise an individual’s visual capacity has not been systematically explored. While our early research used examination times of 30 minutes, we have more recently gathered data over longer repeated examination periods. Clearly, a clinical application would need to minimise examination time and establishing a suitable trade-off between examination duration and data quality needs to be established. Third, validation against other measures is required. We have done this to some extent, but further work is required. Fourth, as for other measures, norms need to be established, a process that is time consuming and may require separate values for each site at which measurements are taken. Fifth, fMRI yields large data sets that need to be processed and then assessed by experts. This process is time consuming compared to the examination time. More work is required on automated procedures to process data. Finally, fMRI is expensive, and cost will always be a consideration and perhaps the determining factor. However, if examination durations can be made short and added to an already required anatomical examination, and if much of the analysis can be automated, then costs will be reduced substantially.

While there are clearly factors holding fMRI back from clinical application, techniques continue to benefit from rapid technological advances that will likely feed through to clinical imaging systems. For example, increasing field strength and the number of coil channels on scanners has increased signal quality. Developments of this type have not hit a ceiling so much more can be expected in the future. Moreover, recent research has shown that novel contrast agents could yield enormous increases in signal over noise.15 Taken together such technological advances will improve the efficiency of fMRI data acquisition, which in turn could lead to faster, cheaper visual testing with fMRI.  

References

1. Holmes G. Disturbances of vision caused by cerebral lesions. British Journal of Ophthalmology 1918;2:353-84.

2. Sereno MI, Dale AM, Reppas JB, et al. Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging [see comments]. Science 1995;268:889-93.

3. Engel SA, Glover GH, Wandell BA. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cereb Cortex 1997;7:181-92.

4. Baseler HA, Brewer AA, Sharpe LT, Morland AB, Jagle H, Wandell BA. Reorganization of human cortical maps caused by inherited photoreceptor abnormalities. Nat Neurosci 2002;5:364-70.

5. Baseler HA, Gouws A, Haak KV, et al. Large-scale remapping of visual cortex is absent in adult humans with macular degeneration. Nat Neurosci 2011;14:649-55.

6. Baseler HA, Gouws A, Crossland MD, et al. Objective Visual Assessment of Anti-Angiogenic Treatment for Wet AMD. Optom Vis Sci 2011;88:X.

7. Baseler HA, Morland AB, Wandell BA. Topographic organization of human visual areas in the absence of input from primary cortex. J Neurosci 1999;19:2619-27.

8. Bridge H, Thomas O, Jbabdi S, Cowey A. Changes in connectivity after visual cortical brain damage underlie altered visual function. Brain 2008;131:1433-44.

9. Meadows JC. Disturbed perception of colours associated with localized cerebral lesions. Brain 1974;97:615-32.

10. Kennard C, Lawden M, Morland AB, Ruddock KH. Color Identification and Color Constancy Are Impaired in a Patient With Incomplete Achromatopsia Associated With Prestriate Cortical-Lesions. Proceedings of the Royal Society of London Series B-Biological Sciences 1995;260:169-75.

11. Hoffmann MB, Tolhurst DJ, Moore AT, Morland AB. Organization of the visual cortex in human albinism. J Neurosci 2003;23:8921-30.

12. Morland AB, Hoffmann MB, Neveu M, Holder GE. Abnormal visual projection in a human albino studied with function magnetic resonance imaging and viusal evoked potentials. J Neurol Neurosurg Psychiatry 2002;72:523-6.

13. von dem Hagen EA, Hoffmann MB, Morland AB. Identifying human albinism: a comparison of VEP and fMRI. Invest Ophthalmol Vis Sci 2008;49:238-49.

14. Morland AB, Baseler HA, Hoffmann MB, Sharpe LT, Wandell BA. Abnormal retinotopic representations in human visual cortex revealed by fMRI. Acta Psychol (Amst) 2001;107:229-47.

15. Adams RW, Aguilar JA, Atkinson KD, et al. Reversible interactions with para-hydrogen enhance NMR sensitivity by polarization transfer. Science 2009;323:1708-11.