Plasticity of visual field representations
De Oliveira Carvalho, Joana
DOI:
10.33612/diss.128352681
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
De Oliveira Carvalho, J. (2020). Plasticity of visual field representations. University of Groningen. https://doi.org/10.33612/diss.128352681
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
CHAPTER
1
Introduction
A key feature of the visual cortex is its topographic organization, whereby neighboring neurons respond to stimulation in nearby regions of the visual field. The portion of the visual field to which a neuron responds is known as its receptive field (RF). When RFs form a representation of the observed image (visual space), this results in a visual field map. The visual system comprises not a single, but many maps of the visual field. These maps are spatially and hierarchically organized.The development of non-invasive imaging techniques, such as functional magnetic res-onance imaging (fMRI), together with biologically inspired neural models, has enabled RFs properties, i.e location and size, to be estimated directly in humans, although at the population level - a level somewhat intermediate between single neurons and that of entire brain areas. In visual neuroscience, these techniques used to delineate the visual field maps in cortex are referred to as visual field mapping or topographic/retinotopic mapping. Ophthalmic or neurological disorders may result in changes in the measured population RFs, for example alterations in their position and/or size. Such changes, also referred to as RF dynamics, have been used to infer the neuroplastic capacity of the adult visual brain. However, there is an ongoing debate whether such population-level changes can be straightforwardly interpreted as an indication of neuroplasticity. This debate has resulted from two recent findings. First, RF dynamics similar to those in pa-tients with visual field defects can be evoked in the neural circuitry of individuals with intact vision, and second, visual field maps of healthy subjects vary substantially, e.g. in response to specific stimuli or under the influence of cognitive factors such as attention. My primary aim with the research presented in this thesis was to enhance our under-standing of how the visual neurons and connecting cortical circuits change, in particular when normal vision is compromised due to a visual field defect or a developmental dis-order. My secondary aim was to develop advanced techniques and paradigms to char-acterize RFs and their connections using neurocomputational models. Accurate tech-niques to do so are required to assess or infer changes in RFs, and thus to increase our understanding of the neuroplastic properties of visual cortex.
1.1
Thesis Outline
My thesis consists of four chapters based on experimental evidence and one review chap-ter. The experiments in Chapters 2, 3 and 5 are focused on the development of new methods and stimuli that result in more accurate measurements of RFs and visual field maps, while those in Chapter 4 are focused on the plasticity and stability of the visual cortex associated with predictive masking (PM) - the process underlying the ability of the visual system to fill-in gaps in perception with the features of nearby regions. Fi-nally, based on these findings and those of others, Chapter 6 presents my view on how to assess visual cortical reorganization.
In chapter 2, we examine to what degree the differences in stimulation may cause changes in measured RFs properties. In the visual cortex, RFs process different spatial and tem-poral features such as orientation, color, luminance, and temtem-poral and spatial frequency. A subpopulation of RFs corresponds to the aggregate of RFs that respond to a specific property. Population receptive field (pRF) mapping captures the cumulative response across all neuronal subpopulations within a voxel (an MRI volume element analogous to a pixel but in 3D space). The conventional stimulus used during retinotopic mapping is a moving high luminance contrast visual pattern (checkerboard like), which drives a response from a wide range of neural subpopulations. While this results in a very strong signal, it also limits the cortical characterization of specific neuronal subpopulations. In Chapter 2we used a second-order stimulus - in which the foreground pattern is delin-eated from the background by differences of contrast or texture, rather than a difference in luminance - to selectively stimulate the orientation sensitive subset of neurons. Our approach resulted in significant differences in position and size relative to the pRF es-timates obtained with the conventional stimulus. This indicates that the recruitment of neural resources depends on the task and/or stimulus.
In chapter 3, we present a new method to assess receptive field properties with great detail. The current approaches to estimate receptive fields (RFs) in humans require a priori assumptions about a voxel’s RFs properties, including its shape and the number of neuronal subpopulations. To overcome this limitation, in Chapter 3, we established and validated a versatile brain mapping technique that captures the activity of neuronal subpopulations with minimal prior assumptions and high resolution, which we called Micro Probing (MP). We used MP to map the bilaterally fragmented receptive fields that are characteristic of congenital visual pathway disorders. Importantly, we did this with-out using specific stimuli or specifically adapted models. In addition, our new technique revealed – also in healthy observers – that voxels can capture the activity of multiple subpopulations of neurons sampling spatially distinct regions of the visual field. In Chapter 4 we unravel the neuronal reconfiguration and cortical circuitry underlying PM. PM plays a very important role in the masking of cortical and retinal lesions. Be-sides being a fundamental attribute of healthy vision, PM has important implications for the early detection of ophthalmologic pathologies. PM hinders the early self-detection of visual field defects, which prevents timely diagnosis and leads to delayed treatment
Chapter
and increased risk of blindness. A marked example of this relates to the ophthalmicpathology glaucoma, in which patients only notice their peripheral visual field defect at an advanced stage. We investigated the neural mechanisms underlying PM. We used fMRI and neural modeling to track changes in cortical population receptive fields (pRFs) and connectivity. We found that in response to the introduction of an artificial scotoma (mimicking a lesion to the visual system), there is a system-wide reconfiguration of cor-tical connectivity and RFs. Based on these findings, we concluded that these changes originate in the lateral occipital cortex (LO), are most likely guided by extrastriate sig-nals, and that they underlie the PM of scotomas.
In Chapter 5 we present a novel method to access visual field loss based on MP using only fMRI. We evaluated the accuracy and reliability of our fMRI-based techniques in healthy participants with simulated scotomas (SS) across different cortical areas (e.g., V1, V2 and V3). In participants with glaucoma, we compared the visual fields obtained with standard automated perimetry (SAP) and fMRI-based techniques. We showed that fMRI-based reconstruction of the visual field enables the evaluation of vision loss and provides important information about the function of the visual cortex. This information may either complement standard techniques to assess the visual field, i.e SAP, or could provide visual field information in patients unable to perform SAP.
Finally, Chapter 6, presents our view on how to study cortical neuroplasticity in oph-thalmic and neurological disorders. We discuss recent advances to improve the reliabil-ity and suitabilreliabil-ity of visual field mapping approaches for studying neuroplasticreliabil-ity. At the same time, we argue that the inherent reliance of current visual field mapping ap-proaches on visual input remains a major challenge for interpreting results in terms of neuroplasticity. Therefore, to assess visual cortical reorganization, we suggest the use of methods that do not require visual stimulation at all, such as cortico-cortical modeling. Such approaches avoid many of the complications associated with the stimulus-driven pRF methods. We also reviewed methods that can be used to do this, including several recent studies that have used such techniques to study plasticity.
1.2
Background
In the following sections, I will briefly describe the organization of a healthy visual sys-tem and how glaucoma affects the visual syssys-tem; I also outline some of the biologically-inspired computational neural models used to detail the structure and connectivity of the visual cortex.
1.2.1
Visual pathways
Light enters the eye through the cornea, and after passing through the pupil it reaches the lens. The optics of the eye (cornea and lens together) projects an inverted image of the outside world onto the retina (layer of photoreceptors: cones and rods), where the light is converted into an electric (neuronal) signal (Dowling and Dowling, 2016).
This signal is transmitted via bipolar cells to ganglion cells. The axons of the ganglion cells exit the eye through the optic disc and form the optic nerve. At the optic chiasm, the nasal fibers from each eye decussate so that the left visual field is processed in the right hemisphere and vice versa (Cotter, 1990). From the optic chiasm, the optic fibers enter the lateral geniculate nucleus (LGN). Postsynaptic fibers leave the LGN as the optic radiations, which terminate in the primary visual cortex (Figure 1.1).
Figure 1.1 Schematic diagram of healthy visual pathways and visual field maps in
hu-mans.The upper row shows the left and right visual fields, which are processed in the
contralat-eral hemisphere after the decussation in the optic chiasm. The bottom row depicts the visual field maps for left and right V1. The image is based on Hoffmann, M. B., Dumoulin, S. O. (2015).
1.2.2
Organization of the Visual cortex
The visual cortex is located in the occipital lobe and consists of a multitude of visual ar-eas. Each visual area is retinotopically organized and processes an entire representation of the contralateral visual space (Wandell et al., 2009). More than 16 visual areas have been identified in the human brain (Wandell et al., 2007). These areas are hierarchically organized. V1 corresponds to the primary visual cortex, as it receives direct input from the LGN and is the first location in the visual pathway where information from the two eyes is combined. V1 is located on the calcarine sulcus and it is bounded by the areas V2 and V3, which consists of a dorsal and a ventral area.
Chapter
Two important characteristics of the visual field maps are the following: 1) the sizeof the receptive fields increases with distance from the center of the visual field, i.e., neurons with relatively small receptive field sizes respond to the central visual field, while neurons with larger receptive field sizes respond to the periphery; and 2) the size of the receptive field increases with visual hierarchy, i.e. neurons in higher-order cortical areas generally have larger receptive field sizes and are therefore less sensitive to spatial location (Dumoulin and Wandell, 2008). The increase of the RF size with visual hierarchy results from the pooling of visual information from V1 towards the temporal and parietal lobes.
Low level features, including spatial frequency, orientation, color, and motion, are pro-cessed by the early visual cortices (such as V1), while increasingly complex features, such as object shape and face, are processed sequentially by extrastriate visual cortices (Grill-Spector et al., 2001; Kanwisher and Yovel, 2006; Malach et al., 1995). The pooling of information from V1 to higher order areas corresponds to the feedforward process-ing of information, which is further complemented with recurrent feedback from higher order areas to early visual areas. It is commonly assumed that the integration of feed-forward and feedback interactions along the visual hierarchy modulates the perception of the visual input (Morgan et al., 2018; Muckli et al., 2015).
1.2.3
Visual field defects
The visual field (VF) corresponds to the complete representation of the visual space, and is usually measured in degrees of visual angle from fixation. The visual field consists of the following components: 1) the fovea, which comprises the central 2.5 degrees from fixation; 2) the central vision, which corresponds approximately to the central 18 degrees diameter of the VF; and 3) the periphery, which is the area of vision outside the central vision.
When a retinal or cortical lesion impairs processing of the complete visual scene, this results in gaps in perception. These are known as visual field defects or scotomas. An example of a natural scotoma is the blind spot, a photoreceptor-less area in the retina where the optic nerve exits the eye (Figure 1.2). In binocular viewing conditions, the blind spot is unnoticed because the corresponding portion of the brain still receives vi-sual input via the other eye. However, the blind spot also remains unnoticed in monoc-ular viewing, because the visual brain fills in the gap. When visual field defects in both eyes affect the same location in the visual field, the corresponding portion of the brain no longer receives stimulation. Also here, the gap may remain perceptually unnoticed due to filling-in. The standard technique to measure the visual field is Standard Automated Perimetry (SAP); a common device for SAP is the Humphrey’s Field Analyzer (HFA). SAP probes the visual field in a systematic way (using a 6x6 degree grid) with a light stimulus and measures the contrast sensitivity in comparison to that of aged-matched controls.
Figure 1.2 HFA printout of a participant with glaucoma (upper row) and a healthy
par-ticipant (lower row).Darker regions correspond to lower contrast sensitivity. Typical contrast
sensitivities are 35 dB for the fovea and 30 dB for the central visual field; black areas correspond to a sensitivity of 0 dB or less. Every -3dB means that the contrast sensitivity is reduced by half. Note the presence of the blind spot in the healthy subject.
1.2.4
Glaucoma
Glaucoma results from a progressive loss of retinal ganglion cells (RGCs) and a thinning of the retinal nerve fiber layer. An increase in intraocular pressure (IOP) is the most important risk factor in glaucoma. Glaucoma is characterized by a gradual increase in visual field defects that typically start in the periphery (Weinreb et al., 2014, A. and H., 1970). Although no cure is known, the management of glaucoma is based on lowering the IOP using drugs (i.e eye drops), or laser therapy and various surgical procedures intended to improve the drainage of fluid from the eye (Beidoe and Mousa, 2012). If left untreated, glaucoma may result in irreversible blindness, so early detection is crucial. However, the disease frequently goes unnoticed as individuals with a scotoma often do not realize that such damage exists until an ophthalmic test is performed. Due to these factors, glaucoma is the leading cause of irreversible vision loss worldwide (Kapetanakis et al., 2016; Tham et al., 2014).
The delay between the onset and the diagnosis of glaucoma results from three factors: 1) the inability of patients to physically perceive high IOP, 2) visual field loss in one eye can be compensated for by information from the other eye, and 3) the brain has the capability to mask the visual field defect, once binocular, by predicting and interpolating
Chapter
the missing information with the visual features (color, brightness, texture, and motion)of the surroundings. The latter is considered the main factor that is responsible for the delay in the early detection of glaucoma.
Damage to the RGC axons, resulting from a pressure difference at the lamina cribrosa, causes the death of the RCG cell bodies and affects the optic nerve (Burgoyne, 2011). This suggests that changes at the level of the brain could also occur. Indeed, mounting evidence indicates that neurodegeneration in glaucoma is not restricted to the retina and optic nerve, but spreads along the entire visual pathway (Chen et al., 2013; Li et al., 2012; Prins et al., 2016; Yu et al., 2013). However, the degree to which these changes affect the visual cortex is highly controversial, especially the notion of functional reorganization resulting from a prolonged lack of visual input.
1.2.5
Predictive Masking (Perceptual filling-in)
When the information extracted from a visual scene is incomplete, the visual system at-tempts to compensate for the gaps in perception by predicting and interpolating based on information from nearby regions of the visual field. This process, known as percep-tual filling-in or predictive masking (PM), is essential for the stability of perception, and it takes place regularly in various forms when people observe their surrounding envi-ronment. For example, PM occurs when objects fall on the blind-spot, when one object is occluded by another, or when people stare steadily for a long time at an image with missing patches of texture (Haak et al., 2015; Weil and Rees, 2011). However, PM is also clinically relevant. Similar to how we are unaware of the blind-spot in our eye when we close the other eye, patients with retinal damage can remain unaware of their defects, especially when the visual field defect does not affect foveal vision and visual acuity is preserved. As a result, PM delays the early detection of lesions and thereby increases the risk of blindness in progressive eye diseases. Despite its scientific and clinical relevance, the neuronal origins of PM are still poorly understood.
According to the most widely supported theory, PM relies on an active neuronal pro-cess rather than on simply ignoring the incomplete portion of the visual field (Komatsu, 2006; Meng et al., 2007; Weil and Rees, 2011). The neuronal activity in the lesion projec-tion zone may result from: 1) lateral propagaprojec-tion of neural signals, with the spread of activation across the retinotopic map of the visual cortex from the border to the interior of the masked surface (De Weerd et al., 1995), and 2) cortical remapping so that the RFs from the lesion projection zone are displaced towards the spared portion of the visual field (Chino et al., 2001). This means that a set of neurons is activated in such a way that a visual stimulus is perceived at a location in the visual field at which no visual input is actually present. But how does this actually happen? Chapter 4 describes our research to understand and model the neuronal processes underlying PM.
1.2.6
Computational Neuroimaging
1.2.6.1 Functional Magnetic Resonance Imaging (fMRI)
Currently, it is not possible to measure the activity of single neurons with non-invasive procedures. However, the development of non-invasive neuroimaging techniques, such as fMRI, has enabled researchers to estimate the neuronal population activity. This approach is possible due to the coupling between blood supply and neuronal activity. Briefly, fMRI uses a magnetic field to detect changes in blood oxygenation, which is used as a proxy for neural activity. Increased neuronal activity elicits a need for oxygen and glucose consumption, which is accompanied by an increase in local blood flow (Logo-thetis and Wandell, 2004). Changes in the ratio between oxyhemoglobin (hemoglobin bound to oxygen) and deoxyhemoglobin (hemoglobin without bound oxygen) can be detected due to their differential magnetic susceptibility and resulting effect on the mea-sured magnetic resonance (MR) signal. The signal intensity will be different depending on whether neurons in a specific part of the cortex were active or not. Areas with high concentrations of oxyhemoglobin (active region) produce a stronger signal than areas with lower oxyhemoglobin levels (inactive region) (Amaro and Barker, 2006). The re-sulting differences between fMRI images enable researchers to distinguish active brain regions (with high oxygen consumption and, consequently, a high influx of oxygen-rich blood) from inactive ones. The dependence of the recorded signal on blood oxygena-tion is the basis of the technique and why it is often referred to as blood-oxygen-level-dependent (BOLD) imaging (Ogawa et al., 1990; Ogawa and Lee, 1990).
1.2.6.2 Non-invasive measurement of receptive fields
The combination of fMRI with biologically-inspired computational neural models has enabled characterization of RF properties at a larger scale. For this purpose, the no-tion of the RF has been extended to the collective RF of a populano-tion of neurons within a voxel, which is called the population Receptive Field (pRF). The technique takes ad-vantage of the retinotopic organization of the visual cortex. A temporally and spatially well-defined stimulus systematically samples every position of the visual field while si-multaneously recording the stimulus-evoked brain activity. This makes it possible to map the topographical and neuroanatomical organization of the human visual cortex. Biologically plausible models such as pRF mapping are now widely used for the detailed characterization of visual cortical maps at the level of neuronal populations (Dumoulin and Wandell, 2008). In essence, this method models the pRF as a two dimensional Gaus-sian, where the center corresponds to the pRF’s position and the width to its size. The model pipeline and description are presented in Figure 1.3.
In the research presented in chapters 4 and 5 we used the pRF model as a means to quan-tify changes in the structure and function of the visual cortex in response to incomplete visual input. We also addressed some of the limitations of this method, and in chapter 3 we proposed a model that enables the characterization of the pRF at a more fine-grained
Chapter
level of detail.Figure 1.3 The pRF modeling procedure.The model describes, per voxel, how the pRF
proper-ties position (x,y) and size (σ) are estimated. The response of a voxel to the stimulus is calculated as the overlap between the stimulus mask (binary image of the stimulus aperture: moving bar) at each time point and the neural model. Subsequently, the delay in hemodynamic response is accounted for by convolving predicted time courses with the hemodynamic response function (HRF). Finally, the pRF model parameters are adjusted for each cortical location to minimize the difference between the prediction and the measured BOLD data. The best fitting parameters are the output of the analysis. Adapted from Dumoulin and Wandell, 2008 (Dumoulin and Wandell, 2008).
1.2.6.3 Cortico-cortical models: Connective Field modeling
To fully understand the plasticity of the visual cortex, it is fundamental to complement the characterization of the RFs with their connectivity. Cortico-cortical models may capture the effects of structural reorganization and can identify which neural circuits have the potential to reorganize and which are stable. An example of this type of model is the connective field (CF) model, which applies the notion of a receptive field to cortico-cortical connections (Haak et al., 2013). In our study, the neuronal response in a target region is predicted based on the response in a source region.
The CF model, as originally proposed by Haak and colleagues (2013), assumes a circu-larly symmetric 2D Gaussian model on the surface of the source region as the integration field from source to target (Haak et al., 2013). This 2D Gaussian is defined by its position (v0)and size (σ); where d(v,v0)is the shortest distance between voxel v and the con-nective field center v0, and (σ) is the Gaussian spread (mm). Distances are calculated across the cortical surface, using Dijkstra’s algorithm (Dijkstra, 1959; Haak et al., 2013). The connective field pipeline is illustrated in Figure 1.4.
B
V1 1 Weight V2dA
V2 pRF model predictionV1>V2 CF model prediction Measured V2 signal
Figure 1.4 A: CF pipeline as described by Haak and colleagues, 2013 (27).The model
com-prises 2 steps: 1) predict the fMRI response, p(t), by multiplying the CF model д(v0, σ) with the
measured source fMRI signal a(v, t) and 2) estimate the CF position (v) and size (σ) by varying parameters and selecting the best fit between the predicted time series and the measured BOLD signal y(t). This procedure is then repeated for every voxel in the target region. B: The V2 re-sponse is predicted based on the pRF (stimulus driven, in blue) and the connective field model (cortical driven, in red). The color map on the brain shows the V1>V2 CF model weights for a specific voxel.
1.3
References
Amaro, E., Jr, Barker, G.J., 2006. Study design in fMRI: basic principles. Brain Cogn. 60, 220–232.
Beidoe, G., Mousa, S.A., 2012. Current primary open-angle glaucoma treatments and future direc-tions. Clin. Ophthalmol. 6, 1699–1707.
Burgoyne, C.F., 2011. A biomechanical paradigm for axonal insult within the optic nerve head in aging and glaucoma. Exp. Eye Res. 93, 120–132.
Chen, W.W., Wang, N., Cai, S., Fang, Z., Yu, M., Wu, Q., Tang, L., Guo, B., Feng, Y., Jonas, J.B., Chen, X., Liu, X., Gong, Q., 2013. Structural brain abnormali-ties in patients with primary open-angle glaucoma: a study with 3T MR imaging. Invest. Ophthalmol. Vis. Sci. 54, 545–554.
Chino, Y., Smith, E.L., 3rd, Zhang, B., Matsuura, K., Mori, T., Kaas, J.H., 2001. Recovery of binocular re-sponses by cortical neurons after early monocular le-sions. Nat. Neurosci. 4, 689–690.
Cotter, J.R., 1990. The Visual Pathway: An Introduc-tion to Structure and OrganizaIntroduc-tion. Science of Vision. 3-15.
De Weerd, P., Gattass, R., Desimone, R., Ungerleider, L.G., 1995. Responses of cells in monkey visual cortex during perceptual filling-in of an artificial scotoma. Nature 377, 731–734.
Dijkstra, E.W., 1959. A note on two problems in con-nexion with graphs. Numer. Math. 1, 269–271. Dowling, J.E., Dowling, J.L., 2016. Vision: How It Works and What Can Go Wrong. MIT Press.
Chapter
Dumoulin, S.O., Wandell, B.A., 2008. Population re-ceptive field estimates in human visual cortex. Neu-roimage. 39, 647–660.
Grill-Spector, K., Kourtzi, Z., Kanwisher, N., 2001. The lateral occipital complex and its role in object recog-nition. Vision Research. 41,1409-22
Haak, K.V., Morland, A.B., Engel, S.A., 2015. Plastic-ity, and Its Limits, in Adult Human Primary Visual Cortex. Multisens Res 28, 297–307.
Haak, K.V., Winawer, J., Harvey, B.M., Renken, R., Du-moulin, S.O., Wandell, B.A., Cornelissen, F.W., 2013. Connective field modeling. Neuroimage 66, 376–384. Kanwisher, N., Yovel, G., 2006. The fusiform face area: a cortical region specialized for the perception of faces. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361, 2109–2128.
Kapetanakis, V.V., Chan, M.P.Y., Foster, P.J., Cook, D.G., Owen, C.G., Rudnicka, A.R., 2016. Global vari-ations and time trends in the prevalence of primary open angle glaucoma (POAG): a systematic review and meta-analysis. Br. J. Ophthalmol. 100, 86–93. Komatsu, H., 2006. The neural mechanisms of per-ceptual filling-in. Nat. Rev. Neurosci. 7, 220–231. Li, C., Cai, P., Shi, L., Lin, Y., Zhang, J., Liu, S., Xie, B., Shi, Y., Yang, H., Li, S., Du, H., Wang, J., 2012. Voxel-based morphometry of the visual-related cor-tex in primary open angle glaucoma. Curr. Eye Res. 37, 794–802.
Logothetis, N.K., Wandell, B.A., 2004. Interpreting the BOLD signal. Annu. Rev. Physiol. 66, 735–769. Malach, R., Reppas, J.B., Benson, R.R., Kwong, K.K., Jiang, H., Kennedy, W.A., Ledden, P.J., Brady, T.J., Rosen, B.R., Tootell, R.B., 1995. Object-related activ-ity revealed by functional magnetic resonance imag-ing in human occipital cortex. Proc. Natl. Acad. Sci. U. S. A. 92, 8135–8139.
Meng, M., Ferneyhough, E., Tong, F., 2007. Dynamics of perceptual filling-in of visual phantoms revealed by binocular rivalry. J. Vis. 7, 8.1–15.
Morgan, A., Petro, L., Muckli, L., 2018. Cortical feed-back to superficial layers of V1 contains predictive scene information. 2018 Conference on Cognitive Computational Neuroscience.
Muckli, L., De Martino, F., Vizioli, L., Petro, L.S., Smith, F.W., Ugurbil, K., Goebel, R., Yacoub, E., 2015. Contextual Feedback to Superficial Layers of V1. Curr. Biol. 25, 2690–2695.
Ogawa, S., Lee, T.M., 1990. Magnetic resonance imag-ing of blood vessels at high fields: in vivo and in vitro measurements and image simulation. Magn. Reson. Med. 16, 9–18.
Ogawa, S., Lee, T.M., Nayak, A.S., Glynn, P., 1990. Oxygenation-sensitive contrast in magnetic reso-nance image of rodent brain at high magnetic fields. Magn. Reson. Med. 14, 68–78.
Prins, D., Hanekamp, S., Cornelissen, F.W., 2016. Structural brain MRI studies in eye diseases: are they clinically relevant? A review of current findings. Acta Ophthalmol. 94, 113–121.
Tham, Y.-C., Li, X., Wong, T.Y., Quigley, H.A., Aung, T., Cheng, C.-Y., 2014. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmol-ogy 121, 2081–2090.
Wandell, B.A., Dumoulin, S.O., Brewer, A.A., 2009. Vi-sual cortex in humans. Encyclopedia of neuroscience 10, 251–257.
Wandell, B.A., Dumoulin, S.O., Brewer, A.A., 2007. Vi-sual field maps in human cortex. Neuron 56, 366–383. Weil, R.S., Rees, G., 2011. A new taxonomy for per-ceptual filling-in. Brain Res. Rev. 67, 40–55. Weinreb, R.N., Aung, T., Medeiros, F.A., 2014. The pathophysiology and treatment of glaucoma: a re-view. JAMA 311, 1901–1911.
Wolter, JR., 1970. System of Ophthalmology: Dis-eases of the Lens and Vitreous; Glaucoma and Hy-potony. Journal of Pediatric Ophthalmology and Stra-bismus.7(2):125.
Yu, L., Xie, B., Yin, X., Liang, M., Evans, A.C., Wang, J., Dai, C., 2013. Reduced cortical thickness in primary open-angle glaucoma and its relationship to the reti-nal nerve fiber layer thickness. PLoS One 8, e73208.
