Comparing Clinical Perimetry and Population Receptive Field Measures in Patients with Choroideremia

Purpose Choroideremia (CHM) is an X-linked recessive form of hereditary retinal degeneration, which, at advanced stages, leaves only small central islands of preserved retinal tissue. Unlike many other retinal diseases, the spared tissue in CHM supports excellent central vision and stable fixation. Such spared topography in CHM presents an ideal platform to explore the relationship between preserved central retinal structure and the retinotopic organization of visual cortex by using functional magnetic resonance imaging (fMRI). Methods fMRI was conducted in four participants with CHM and four healthy control participants while they viewed drifting contrast pattern stimuli monocularly. A single ∼3-minute fMRI run was collected for each eye separately. fMRI data were analyzed using the population receptive field (pRF) modeling approach. Participants also underwent ophthalmic evaluations of visual acuity and static automatic perimetry. Results The spatial distribution and strength of pRF estimates correlated positively and significantly with clinical outcome measures in most participants with CHM. Importantly, the positive relationship between clinical and pRF measurements increased with increasing disease progression. A less consistent relationship was observed for control participants. Conclusions Although reflecting only a small sample size, clinical evaluations of visual function in participants with CHM were well characterized by the spatial distribution and strength of pRF estimates by using a single ∼3-minute fMRI experiment. fMRI data analyzed with pRF modeling may be an efficient and objective outcome measure to complement current ophthalmic evaluations. Specifically, pRF modeling may be a feasible approach for evaluating the impact of interventions to restore visual function.

sensitivity of an individual as compared to the visual sensitivity of a normal age-matched control.
In these plots, bright-colors (yellow/white) represent locations in the visual field that are either as sensitive or better than a normal age-matched control, whereas dark-colors (red/maroon) represent location in the visual field with decreased sensitivity from a normal age-matched control. As expected, all HCs demonstrate normal or near normal visual sensitivity across the tested visual field. In each HC, the majority of the SAP maps are bright, which demonstrates similar sensitivity to a normal age-matched control. In contrast, the dramatic reduction in peripheral sensitivity is clearly evident for CHM participants (P2-P4), with dark-colored cells (poor visual sensitivity) in the periphery. For example, P4 who was at a late disease stage exhibits only moderately bright cells at the very center of the visual field for both the left and right eyes. P1, who represents early stages of disease progression exhibits relatively normal visual field sensitivity, particularly for the left eye. In the right eye, the fair right visual field has poorer sensitivity. Figure 1. Total deviation SAP maps for each CHM participant, HC and eye, respectively. Total deviation maps represent the level of visual sensitivity relative to a normal age-matched control. In these plots, bright-cells (yellow/white) represent locations in the visual field with sensitivity either equal or greater than normal age-matched controls, whereas darkcolors (red/maroon) represent locations in the visual field with decreased sensitivity relative to a normal age-matched control. The black-line on the color-bar represents the same sensitivity as a normal age-matched control. Thus, bright yellow/white cells indicate increased sensitivity and orange/red/maroon cells indicate decreased sensitivity.

Supplementary Figure 2.
Right eye pRF visual field coverage plots for four CHM participants and one healthy control (HC1). Similar to the left eye, assessment of the visual fields was restricted to + 10 and -10 degrees of the entire visual field. The pRF distribution for a typical healthy control (HC1) is presented in the center image. As expected, similar to the left eye, HC1 presented with high pRF values across the visual field with relatively high intensity levels except at the extreme upper and lower vertical meridians, a pattern regularly observed in pRF mapping.
Coverage plots for CHM participants' right eye were assessed relative to HC1's maximum pRF intensity levels. pRF distribution for the CHM participant with most vision (P1) showed similarity to HC1 with higher intensities in the upper left visual field but great reduction of visual intensities in the lower left quadrant. The CHM participant with more advanced disease and limited visual field (P4) showed small centrally located pRF distribution with moderate intensity values. The two CHM participants in mid disease stages (P2 and P3) both showed dramatically more attenuated visual intensities for their right eye pRF distributions as compared to HC1. At the same token, their pRF distributions also differed dramatically from each other. While P2 presented with increase central and right peripheral visual field, P3's visual field coverage was mainly limited to a strong central representation, which was largely restricted to the upper visual field.

Quantification of Relationship Between SAP and pRF Measurements
To quantify the relationship between SAP measurements and pRF parameters, visual field coverage maps needed to be resampled to match the spatial sampling of SAP measurements.
Thus, based on the SAP of a 10-2 (10x10 matrix), visual field coverage maps were divided evenly into 10x10 square grids, with each grid representing 2x2 degrees of visual angle

Quantification of Relationship Between Clinical and pRF Horizontal Eccentricity Profiles
To show the high predictability of the pRF measures for CHM participants' eccentricity profiles, the visual field coverage maps needed to be converted to represent a similar spatial distribution

Eccentricity Measurements Across the Cortical Surface
Although the majority of analyses were conducted on the significantly modulated voxels pooled across both hemispheres, it is also noteworthy to demonstrate the distribution of pRF eccentricity across the cortical surface. The cortical representation of eccentricity, derived from the right eye pRF scans, are shown on surface reconstructions of both left and right hemispheres for a representative HC (HC1), and two CHM patients at their early and late disease stages respectively (P1 and P4), (see Supplementary Figure 6). As expected, the cortical distribution of eccentricity in HC1 follows a well described and predictable pattern 16 for a typical healthy sighted control, extending gradually from the occipital pole, representing central vision, to representations of the peripheral visual field more anteriorly (top row Supplementary   Figure 6). Similar to HC1, the cortical distribution of eccentricity in P1 (early stage) shows a general posterior to anterior progression from foveal to peripheral, although the eccentricity distribution in P1 appears to be shifted posteriorly relative to HC1 (middle row Supplementary   Figure 6). In stark contrast to both HC1 and P1, the cortical distribution of eccentricity in P4 (late stage) is almost exclusively foveal and is shifted anteriorly (bottom row Supplementary Figure   6). Figure 6. Right eye pRF eccentricity distribution depicted on medial aspects of the left and right inflated hemispheres (dark and light gray colors represent sulci and gyri respectively) for a healthy control (HC1) and two CHM patients, P1 and P4 who are at early and late disease stages respectively. To best demonstrate differences between the HC1 and CHM patients, spatial distribution of eccentricity from fovea (Central position = 0) to the peripheral visual field (peripheral location = 10 degrees) are overlaid using false color where dark blue represents foveal vision, red represents the peripheral visual field at 10 degrees and light blue, yellow and green represents the visual spaces between 0-10 degrees. As anticipated, the representation of eccentricity in HC1 progresses gradually from the fovea (dark blue) at the occipital pole, towards the representation of the periphery (red) more anteriorly. The spatial distribution of eccentricity in P1 with early stage disease also shows a general progression from foveal posteriorly towards more peripheral anteriorly, although relative to HC1, the visual field is primarily limited to central visual field (blue colors) with a more restricted peripheral representation that is considerably shifted posteriorly. In addition, there is far less representation of the periphery dorsally than ventrally. In contrast to both HC1 and P1, the spatial distribution of eccentricity for P4 is strikingly different. Almost all of the cortical representation of the visual field is limited to foveal vision (blue) with an overall anterior shift

Alternative pRF Implementations
Although all pRF models follow a similar general approach, some differences exist. For instance, the pRF implementation used in the current study applies a linear model, similar to the original pRF description 20 , but more recent pRF implementations include compressive spatial nonlinearities 22 and deviations from strictly circular pRF shapes 21 . Data analyzed with these different implementations may produce subtlety different pRF estimates, although it is unlikely that these subtle differences would change the overall pattern of results reported here, namely the strong positive relationship reported between the pRF and clinically measured data for CHM participants. Nevertheless, future studies could test the effect that different pRF implementations have on the correspondence between clinical and pRF measurements.
It is also worth noting that although CHM participants and controls performed a task at the fixation point, they were not performing a task on the stimulus explicitly. A number of previous studies have investigated the effect of stimulus task on fMRI signals in visual cortex of patients with retinitis pigmentosa 28 and macular degeneration 42 , respectively. It is possible that performing a task on the stimulus 43 may alter the patterns of activity in the periphery of participants with CHM and future studies could explore this possibility.

Reliability of pRF estimates
Unfortunately, in the current CHM and HC groups only a single ~3-minute pRF run was acquired for each eye separately, which makes quantification of pRF reliability difficult. Ideally multiple runs for each eye would be acquired in order to demonstrate the reliability and robustness of the pRF estimates that this model derives. However, previous work 31 employing the exact computational model used here, has reported the reliability of the pRF estimates across independent sets of data in healthy participants. This report, demonstrates high degree of reliability (0.68 < r < 0.94) across the three main parameters of interest and provides reasonable confidence with respect to the pRF measurements acquired in the current experiment.