Organization of area hV5/MT+ in subjects with homonymous visual field defects

Abstract

Damage to the primary visual cortex (V1) leads to a visual field loss (scotoma) in the retinotopically corresponding part of the visual field. Nonetheless, a small amount of residual visual sensitivity persists within the blind field. This residual capacity has been linked to activity observed in the middle temporal area complex (V5/MT+). However, it remains unknown whether the organization of hV5/MT+ changes following early visual cortical lesions. We studied the organization of area hV5/MT+ of five patients with dense homonymous defects in a quadrant of the visual field as a result of partial V1+ or optic radiation lesions. To do so, we developed a new method, which models the boundaries of population receptive fields directly from the BOLD signal of each voxel in the visual cortex. We found responses in hV5/MT+ arising inside the scotoma for all patients and identified two possible sources of activation: 1) responses might originate from partially lesioned parts of area V1 corresponding to the scotoma, and 2) responses can also originate independent of area V1 input suggesting the existence of functional V1-bypassing pathways. Apparently, visually driven activity observed in hV5/MT+ is not sufficient to mediate conscious vision. More surprisingly, visually driven activity in corresponding regions of V1 and early extrastriate areas including hV5/MT+ did not guarantee visual perception in the group of patients with post-geniculate lesions that we examined. This suggests that the fine coordination of visual activity patterns across visual areas may be an important determinant of whether visual perception persists following visual cortical lesions.

Keywords: Blindsight; Cortical lesion; Reorganization; Visual cortex; fMRI.

Fig. 1.. hV5/MT+ visualfield coverage maps in Artificial Scotoma subjects.

Visual field coverage density maps of area hV5/MT+ of the right hemisphere in all control subjects with an AS at the upper left quadrant using a direct-fit pRF method (A), a topography-based method (B) and the proposed pRF boundary mapping method (C). The color map indicates the number of pRFs that cover each visual field location. The pRF centers from all voxels are plotted as grey dots. For both direct-fit and topography-based pRF methods, hV5/MT+ maps cover significantly the area of the AS at the upper left quadrant when the full bar stimulus is used for modeling the pRFs. In contrast, for the proposed method (C) in which we map the pRF boundaries directly from the BOLD signal, pRFs in hV5/MT+ are confined to the lower visual field quadrant outside of the AS in all subjects. Activity only modestly crosses the border of the AS (~2 deg) commensurate with the subject’s fixation eye movements. Note that the boundary mapping pRF method is more strict than the direct-fit and the topography-based methods as only well-defined pRFs are retained for analysis. Nevertheless, applying a more strict criterion (variance explained > 30%) for the direct-fit and topography-based method does not eliminate the pRF biases observed within the AS.

Fig. 2.. Anatomical location of the lesion and visual field perimetry tests.

A. Anatomical location of the lesion. A sagittal and an axial slice illustrates each patient’s anatomical lesion (a red arrow points to the lesion). B. Pattern deviation probability plots of the 10-degree Humphrey type (10–2) visual field test for all patients. The small black dots show the locations in the visual field that are normal, while the black squares indicate a visual field defect on a p<0.5% level according to the pattern probability plot (this means that less than 0.5% of normal subjects would be expected to have such a low sensitivity at this visual field location). Pattern deviation numeric plots for all patients had visual sensitivity < −20 dB (absolute visual field scotoma) at all visual field locations within the affected quadrants. Black square locations outside the affected quadrants showed visual sensitivity < −10 dB (mostly still < −20 dB). C. Binocular semi-automated 90° kinetic perimetry (Octopus 101; methods) for patients P1, P2, P4 and P5. The area of absolute visual field loss is shaded in light grey.

Fig. 3.. Visual field coverage maps and average BOLD signal change in hV5/MT+ of the lesioned hemisphere.

A. Left: Pattern deviation probability plots of the 10-degree Humphrey type visual field test for all patients as shown in Fig. 2B. Right: Visual field coverage density maps of area hV5/MT+ of the right hemisphere of an AS control subject (a) and of the lesioned hemisphere for each patient obtained using the proposed pRF boundary mapping method (b–f). To allow comparison between subjects, the scale of the color map has been clipped to the average significantly activated number of voxels of hV5/MT+ of AS controls (97.8 ± 89.15). The total number of significantly activated voxels in hV5/MT+ of each subject is indicated next to the graphs with a # symbol. The pRF centers from all voxels within each area are plotted as grey dots. The average coverage density map of all AS controls is overlaid on top of the maps of each patient with magenta color. In contrast to the AS controls, the visual field coverage maps of hV5/MT+ of all patients cover areas that overlap with the dense visual field scotoma (red arrows). B. The average BOLD signal change from all voxels in the right hV5/MT+ in controls and the hV5/MT+ of the lesioned hemisphere in patients as a horizontal bar is moving from the top (elevation>0; AS/scotoma) to the bottom of the visual field (elevation<0; seeing quadrant). Before averaging, the BOLD time series of each voxel is deconvolved to remove the hemodynamic response function (Methods) and the baseline is removed. The baseline here is defined as the signal value when the vertical bar is located in the far ipsilesional part of the visual field, which should produce little or no visual modulation in the region examined. This is calculated as the average BOLD signal change over 5 steps of the bar when the horizontal bar was located between 7 and 10° in the hemifield ipsilateral to our ROI. This procedure sets the baseline of each voxel to zero. (a) The average signal of the AS controls (white bars) is compared with the full field stimulus condition (blue bars). When the AS is applied, the average BOLD signal change when the bar is in the superior quadrant (location of the AS; elevation>0) drops to baseline values compared with the average signal under the full field stimulus condition. Activity starts when the bar is near 2° from the horizontal meridian (AS border), commensurate with the subject’s fixation eye movements. (b–f) The average signal of the patients (grey bars) compared with the AS controls (white bars). For all patients, activity starts when the stimulus is located well within the perceptual scotoma (elevation>3°, red arrows) in contrast to the AS controls. The error bars indicate the standard error of the mean across control subjects (N = 5). The snapshot on top shows the orientation of the bar and direction of motion (white arrow). C. Same as in (B), the average BOLD signal change from all voxels in the right hV5/MT+ in controls and the hV5/MT+ of the lesioned hemisphere in patients as a vertical bar is moving from the contralateral (azimuth>0) to the ipsilateral visual hemifield (azimuth<0). For all patients, activity starts when the stimulus is located about 2–3° within the ipsilateral visual hemifield, similar to the AS control subjects. This suggests that, hV5/MT+ activity in the lesioned hemisphere originates from stimulus positions located within the scotoma, and less likely from the contralateral hemisphere.

Fig. 4.. Visual field coverage maps of V1 and hV5/MT+ in Control Subjects.

Top: Visual field coverage density maps of area V1 (A) and the hV5/MT+ complex (B) of the right hemisphere of a control subject under Full Field (FF) stimulation. The pRF centers from all voxels within each area are plotted as grey dots. Both areas cover the entire left visual hemifield, as expected. Bottom: Visual field coverage density maps of V1 (A) and hV5/MT+ (B) of the right hemisphere of a control subject under the Artificial Scotoma (AS) condition. No activity is observed within the area of the AS in control subjects for both V1 and hV5/MT+. C. Visual field coverage maps of hV5/MT+ based on the connective field modeling method (CFM) for a control subject under FF stimulation (top) and under the AS condition (bottom). These maps plot, for each voxel in hV5/MT+, the pRFs of the voxels corresponding to the CF center and CF Gaussian spread in the cortical surface of V1. The color map indicates the CF weight so that 1 corresponds to the CF center. CFM coverage maps in control subjects are commensurate to the pRF coverage density maps suggesting that hV5/MT+ activity arises chiefly from area V1.

Fig. 5.. Visual field coverage maps of patients P2, P3 and P5.

A. Pattern deviation probability plots of the 10-degree Humphrey type visual field test for patients P2, P3 and P5 as shown in Fig. 2B. B. Visual field coverage density maps of the spared part of area V1 and C. the hV5/MT+ complex of the lesioned hemisphere for patients P2, P3 and P5. The scale of the color map has been clipped to the average significantly activated number of voxels plus one standard deviation for V1 (182.4 ± 182.3) and the average significantly activated number of voxels for hV5/MT+ (97.8 ± 89.15) of AS controls. The total number of significantly activated voxels for each subject is indicated next to the graphs with a # symbol. The pRF centers from all voxels within each area are plotted as grey dots. The average coverage density map of all AS controls is overlaid on top of the maps of each patient with magenta color. For these patients activity extending beyond the border of the scotoma in hV5/MT+ is also present in V1 (red arrows). Apparently, this activity is not sufficient to mediate conscious vision suggesting it is either too disorganized to elicit a percept or that damage to other areas is responsible for the visual deficit. D. Visual field coverage maps of hV5/MT+ based on the connective field modeling method (CFM). CF coverage maps in patients P2, P3 and P5 cover visual field locations overlapping with the scotoma confirming that responses in hV5/MT+ within the scotoma arise from the spared part of area V1.

Fig. 6.. Visual field coverage maps of patients P1 and P4.

A. Pattern deviation probability plots of the 10-degree Humphrey type visual field test for patients P1 and P4. B. Visual field coverage density maps of the spared part of area V1 and C. the hV5/MT+ complex of the lesioned hemisphere for patients P1 and P4. The pRF centers from all voxels within each area are plotted as grey dots. The average coverage density map of all AS controls is overlaid on top of the maps of each patient with magenta color. The total number of significantly activated voxels for each subject is indicated next to the graphs with a # symbol. Patients P1 and P4 have visual field areas overlapping with the patients’ scotoma that are covered by V5/MT+ but not V1 (green arrows) suggesting the existence of functional V1-bypassing pathways. Activity in hV5/MT+ alone is not sufficient to elicit a percept. Patient P1 also has a part of the visual field overlapping with the patient’s scotoma that is covered by both areas V1 and hV5/MT+ similar to patients P2, P3 and P5. D. Visual field coverage maps of hV5/MT+ based on the connective field modeling method (CFM). For patients P1 and P4, the CF modeling links the voxels in hV5/MT+ to voxels in V1 with pRF centers in the inferior (seeing) quadrant only suggesting that responses observed within the scotoma in hV5/MT+ using the pRF mapping method are independent of V1 input.

Fig. 7.. Cortico-cortical connectivity between hV5/MT+ and V1 in patients.

Top row, red: The location of voxels in hV5/MT+ with pRF center elevation in the inferior (seeing) quadrant (y < 0) is plotted as a function of the pRF center elevation of the corresponding CF center in V1 (as found using the CF modeling method, see methods). The grey shaded area represents the mean ± standard deviation of the AS controls (N = 5). Bottom row, blue: The location of voxels in hV5/MT+ with pRF center elevation in the superior quadrant (scotoma, y > 0) is plotted in blue as a function of the pRF center elevation of the corresponding CF center in V1. The error bars indicate the standard deviation across voxels within an elevation bin (bin size = 0.5°) for each patient. For patients P1 and P4, voxels in hV5/MT+ that have pRF centers within the scotoma (y > 0; superior quadrant) are linked (in the connective field sense) only with voxels in V1 that have pRF centers in the inferior (seeing) quadrant. This correspondence is retinotopically ectopic confirming that the retinotopically corresponding V1 voxels have been lesioned. This ectopic association suggests that visually driven hV5/MT+ responses within the scotoma do not have their source in visual responses of surviving V1 voxels, but instead arise through V1-bypassing pathways. Note that in the control subjects with the artificial scotoma this situation does not arise. For patients P2 and P3, voxels in hV5/MT+ that have pRF centers within the scotoma (y > 0) are linked with voxels in V1 whose pRF centers belong to either the superior (scotoma) quadrant or the inferior (seeing) quadrant, suggesting that hV5/MT+ responses within the scotoma may arise within the spared V1 cortex, through V1-bypassing pathways, or via a combination of both.

Fig. 8.. Population receptive field size in hV5/MT+ of the lesioned hemisphere.

A. Histograms of the distribution of pRF size from hV5/MT+ of all patients (grey bars) compared with the mean distribution of AS controls (step histogram). The shaded area indicates the SEM across the AS controls. The pRF size distribution of patients P1, P3 and P5 is shifted toward larger pRF sizes compared with the AS controls. The mean and standard deviation of each distribution for each patient is indicated on top of the graphs. The mean and standard deviation of the average distribution of AS controls is indicated in blue color. B. Same as in (A) but for pRF centers that are located in the inferior (seeing) quadrant only (pRF elevation < 0).

Fig. 9.. Visual field coverage maps and population receptive field size in hV5/MT+ of the contra-lesional hemisphere.

A. Visual field coverage density maps of hV5/MT+ of the contra-lesional hemisphere for all patients. The scale of the color map has been clipped to the average significantly activated number of voxels for hV5/MT+ (218 ± 104) of AS controls. The total number of significantly activated voxels for each subject is indicated on top of the graphs with a # symbol. The pRF centers from all voxels within each area are plotted as grey dots. B. Histograms of the distribution of pRF size from hV5/MT+ of the contra-lesional hemisphere of all patients (grey bars) compared with the mean distribution of the hemisphere ipsilateral to the AS (left) of AS controls (step histogram). The shaded area indicates the SEM across the AS controls. The mean and standard deviation of each distribution for each patient is indicated on top of the graphs. The mean and standard deviation of the average distribution of AS controls is indicated in blue color.