Development of a wide-view visual presentation system for visual retinotopic mapping during functional MRI

Authors

  • Tianyi Yan PhD,

    1. Biomedical Engineering Laboratory, The Graduate School of Natural Science and Technology, Okayama University, Okayama, Japan
    Search for more papers by this author
  • Fengzhe Jin PhD,

    1. Graduate School of Engineering, Kagawa University, Japan
    Search for more papers by this author
  • Jiping He PhD,

    1. Center for Neural Interface Design, Arizona State University, Tempe, AZ USA, and Huazhong University of Science and Technology, Wuhan, China
    Search for more papers by this author
  • Jinglong Wu PhD

    Corresponding author
    1. Biomedical Engineering Laboratory, The Graduate School of Natural Science and Technology, Okayama University, Okayama, Japan
    • Biomedical Engineering Laboratory, The Graduate School of Natural Science and Technology, Okayama University, Okayama, Japan, 1-1 Tsushima- naka, 3-Chome, Okayama 700-8530, Japan
    Search for more papers by this author

Abstract

Purpose

To develop and validate the functionality of a novel wide-view visual presentation system with a horizontal and vertical eccentricity angle of 60° for retinotopic mapping by functional MRI (fMRI).

Materials and Methods

The wide-view presentation system consisted of a 52-mm diameter optical fiber, an entrance apparatus and a presentation apparatus. The terminal edge of the optical fiber at the entrance is flat, while the terminal edge on the presentation apparatus is a sphere of 60 mm in diameter. The subjects wore contact lenses with +20, +22, or +25 magnification to focus on the stimulus, and the visual field eccentricity angle could reach 60°. The signal to noise ratio valuation experiment was performed to evaluate the clarity and quality of the MRI picture image. Checkerboard and random dot stimuli were used to prove that this system could be applied to retinotopic mapping by fMRI.

Results

The results of the experiment demonstrated that the system is safe in the MRI environment with minimal distortion and can be used for visual retinotopic mapping studies. Wide-field mapping areas (V6, MST) were found in the human visual cortex. Compared with previous studies, the V1 and MT+ surface area approaches but does not fully cover the anatomical area. Nonetheless, the area achieved using the new system is larger than those achievable in previous fMRI studies.

Conclusion

We developed a versatile, low-cost system for presenting wide-view visual stimuli in the MRI environment. The fMRI retinotopic mapping results proved the viability of this system. J. Magn. Reson. Imaging 2011. © 2011 Wiley-Liss, Inc.

OVER THE PAST 10 years, human visual area mapping has been widely investigated using functional magnetic resonance imaging (fMRI) (1, 2). The results have enabled us to study, noninvasively and precisely, the human visual processing characteristics, but the mode of presenting visual stimuli remains a major problem in fMRI studies. Various visual stimuli presentation systems have been developed in the past. The most common method is the back-projection, which uses an LCD projector to project stimuli onto a rear-projection screen. Subjects look at stimuli on an adjustable angled mirror (usually mounted on the head coil of the MRI), reflected from the projection screen (3). Another approach by Cornelissen et al. and Brewer et al. used optic fibers to relay stimuli from an ordinary CRT display to the eyes of a subject lying in the scanner (4, 5), but they were limited in presenting wider-view stimuli.

Retinotopic mapping is one of the most important experimental procedures in human visual cortical research (6). However, due to the technical limitations of visual presentation, previous retinotopic studies on human visual retinotopic mapping are typically limited to central and/or peri-central (8–30° of eccentricity from the fixation point) visual fields (7–9). Visual abilities change over the visual field. A previous study showed that the ability to detect movement is better in peripheral vision than in foveal vision, but color discrimination is markedly worse (10, 11). To reveal the character of the visual retinotopic mapping for the peripheral visual area, many anatomical studies have examined the periphery visual field in primates. For example, Adams studied the peripheral vision (0–72° of eccentricity) characteristics of monkeys using angioscotoma mapping (12, 13). In humans, investigation of peripheral visual cortex characteristics has been conducted mainly through anatomical and physiological postmortem studies (14, 15).

To study retinotopic mapping for human peripheral visual fields using neuroimaging, Cheng et al used smaller displays and lenses to achieve visual fields reaching up to 40° of eccentricity (16). However, this system was difficult to use for fMRI studies due to the restricted space of the fMRI environment. Recently, Pitzalis et al placed a screen/mirror very close to the subject, reaching 55° of eccentricity (horizontal meridian approximately 55° of eccentricity, vertical meridian approximately 40° of eccentricity) (20, 26). The method required a special head coil that is extremely costly for use in general clinical MRI systems. In addition, this system presents stimuli that are not isotropic because the space for the bore of the scanner is very small and the aspect ratio of the mirror varies. This greatly restricts human visual cortical research using fMRI.

The aim of the present study is to develop a wide-view visual presentation system with a horizontal and vertical visual angle of 120° in MRI environment for vision research and visual retinotopic mapping. We will present the system design and the preliminary experimental testing of the system in this report. The experimental results suggest that the new system is safe and functional in the MRI environment and that it can be used for neuroimaging studies of the visual system. In addition, we also located separate peripheral visual field representation areas (V1, V2, V3) and verified recent findings that human motion areas (MT+) are apparent only when wide-view visual stimuli are used. The results prove that the wide-view visual presentation system can be used to map other visual areas with well-developed peripheral representations.

MATERIALS AND METHODS

Overview of Wide-View Visual Presentation System

As MRI involves the use of strong magnetic fields around the head of a subject within a narrow tunnel, any device used during MRI is required to be free of ferromagnetic elements and should not interact with the magnetic field (Fig. 1). Therefore, we used nonmagnetic materials to build the device and optical fiber bundle to present visual stimulus patterns. The wide-view visual stimulus presentation system consists of a computer, a visual stimuli entrance apparatus, an optical fiber bundle, and a visual stimuli presentation apparatus (Fig. 1). The computer was used to control the presentation of stimuli by the operator in the operation room.

Figure 1.

a: The wide-view visual stimulus presentation system. The system consists of four parts: operation computer, optical fiber bundle, visual stimuli entrance apparatus, and visual stimuli presentation apparatus. The visual stimuli presentation apparatus contains the stimulation transmission optical fiber bundles and the structure used to fix the fiber bundles. The right graph is the enlarged drawing of the entrance apparatus, which is marked on the apparatus. b: The visual stimuli presentation apparatus stand in the fMRI scanner head coil. The forehead rest protects the observer's eyes from contact with the eyepieces. The distance is 30 mm from the eye to the center of the optic fiber bundles. The edge side in the presentation part of the optical fiber is a sphere surface of 60 mm in diameter. c: The entrance apparatus consists of an entrance display, a section of optical fiber bundles, the optic fiber bundle supporting stand and display box. Visual stimuli were created on a displayer using a resolution of 800 × 600 pixels. d: The optical fiber bundle is composed of optical fibers 0.5 mm in diameter; the bundle diameter is 52 mm, and the length is approximately 5500 mm. The edge side in the entrance part of the optical fiber is a flat surface of 52 mm in diameter.

Visual Stimuli Presentation Apparatus

The visual stimulus presentation apparatus includes a section of the optical fiber bundle and a holder. Inside the scanner, a plastic optical bundle holder rests on the head coil and supports the optical bundle. The holder is made of nonferrous materials (PVC) and is easy to set up and remove. It also allows easy adjustment of the optic bundle position to accommodate various head sizes, eye positions, and inter-pupillary distances. Monocular (right eye) presentations were performed with the optical-fiber screen (with a curved surface having a 30-mm curvature radius) placed in the center of the bore, 30 mm from the subjects' eyes. The visual field of the stimulus was 120° horizontal × 120° vertical. Because the screen was so close to the eyes, subjects wore contact lenses (Menicon soft MA; Menicon, Japan) with +20, +22, or +25 magnification to retain their length of focus (Fig. 1). The enlargement was taken into account when the sizes of the stimuli were calculated. Refraction of light in the three diopter lenses increased the stimulus size by approximately 8.9% (+20) to 11.3% (+25) of the stimuli (Fig. 1B).

Visual Stimuli Entrance Apparatus

The entrance apparatus consists of an entrance display, a section of the optical fiber bundle, a supporting stand for the fiber bundle and a display box (Fig. 1). Visual stimuli were created on a display with a resolution of 800 × 600 pixels (Fig. 1C).

Optical Fiber Bundle

The optical fiber bundle is composed of 8700 optical fibers (0.5 mm in diameter, CK-20, Eska; Mitsubishi, Japan) for a total outer bundle diameter of 52 mm, and the length approximately 5500 mm. In the center portion, optical fibers were arranged so that the stimuli could be presented in the same way as presented at the entrance. The edge side in the entrance part of the optical fiber bundle is a planar surface, whereas the edge side of the presentation part is a sphere of 60 mm in diameter. To make the presentation part into a sphere of 60 mm in diameter, we bonded the presentation part by using soap bubbles of 60 mm in diameter while performing a positional match at each spatial position. To prevent light from mixing into the plasmotomy and the outside of the optical fiber with the visual stimulus, the optical fiber bundle was covered with a plastic sleeve (Fig. 1D).

Safety and Signal to Noise Ratio Tests

The display in the entrance apparatus located in the MRI room may interfere with MRI images. Therefore, we set up a test to evaluate safety and image quality using a phantom. The wide-view presentation system is presented with the retinotopic stimulus during MR scan (spin echo, repetition time/echo time [TR/TE] = 3000/15 ms, 256 × 256 matrix, 15 continuous 5-mm slices without gap).

Retinotopic Mapping by Wide-View Presentation System

Eight healthy subjects (age, 19–31 years; mean, 25 years; two women, six men) without previous neurological or psychiatric disorders participated in the study. The subjects had normal or corrected-to-normal vision and were all right-handed. The related technical support and contact lenses are provided by the professional company Menicon in Japan. Each subject wore contact lenses with the conduction of the specialized ophthalmologist. The study was approved by the Institutional Research Review Board of Kagawa University, Japan.

To identify the retinotopic areas of the visual cortex, fMRI scans were carried out while the subjects viewed phase-encoding stimuli (1). For each hemi-field, two types of stimuli were used to locate visual area boundaries and to estimate the eccentricity. The stimulus for locating boundaries was a 22.5° wedge that rotated slowly counterclockwise around a red fixation spot. The wedge rotated at steps of 22.5°, remaining at each position for 4 seconds before rotating to the next. The stimulus for estimating eccentricity was an expanding checkered annulus. The flickering radial checkerboard was moved from the center to the periphery in discrete steps (each step 7.5°, with a total of four steps), remaining at each position for 8 s before automatically expanding to the next position (18).

The stimulus for locating area MT+ was functionally identified based on responses to stimuli that alternated in time between moving and stationary dot patterns (Fig. 4, using methods described by Huk et al, 19). We also determined the area MT+ (MT, MST) by a similar moving stationary dot patterns. Moving dots traveled toward and away from fixation (8 °/s) within a 120° diameter circular aperture, alternating direction once per second. To compare contralateral and ipsilateral responses in the MT and MST areas, we tested ipsilateral responses using stimuli restricted to either the left or the right hemifield (Fig. 4). The dots were restricted to a peripheral circular aperture (120° diameter) with its closest edge 15° from fixation.

MR Data Acquisition

The fMRI experiment was performed using a 1.5 Tesla (T; Philips clinical scanner (Intera Achieva; Best, The Netherlands). All images were acquired using a standard radiofrequency head coil. We acquired 23 slices approximately orthogonal to the calcarine sulcus to cover most of the cortical visual areas. The T2*-weighted gradient echo-planner imaging sequence was used with the following parameters: TR/TE = 2000/50 ms; FA = 90°; matrix size = 64 × 64; and voxel size = 3 × 3 × 3 mm. Before acquiring the functional images, T2-weighted anatomical images were obtained in the same planes as the functional images using the spin echo sequence. A T1-weighted high-resolution image (1 × 1 × 1 mm) was also acquired after each functional experiment.

Data Analysis

The functional and anatomical data were processed using BrainVoyager software (Brain Innovation, Masstricht, Netherlands). The recorded high-resolution T1-weighted three-dimensional (3D) scans were used for surface reconstruction. The gray and white matter was segmented using a region-growing method, and the white matter cortical surface was reconstructed. Before surface flattening, the cortical surface was inflated and cut along the calcarine from the occipital pole to slightly anterior to the parieto-occipital sulcus.

The functional data were aligned onto the 3D anatomic image using the image coordinates. To identify boundaries (wedge stimuli), maps were created based on cross-correlation values for each voxel, as determined by a standard hemodynamic box-car function (r ≥ 0.25).

RESULTS

Safety and Signal to Noise Ratio Tests

The signal to noise (SNR) was calculated using the following formula [1]:

equation image(1)

Sp is mean of the signals in the phantom; Nair is the standard deviation of outside noise (Fig. 2). The average SNR calculated from ten MRI images for two conditions, respectively, is shown in Figure 2. The SNR results with and without the wide-view visual presentation system were 110.85 and 117.22. The digressive rate of image with the device was 6.37%. We found no artifacts created by radiofrequency noise in the phantom's phase encoding.

Figure 2.

a,b: Acquired image of the spherical phantom, when the wide-view visual presentation system was not installed (a), and when the system was operated (b).

Delineation of Retinotopic Areas by Wide-View Stimuli

Figure 3 shows a 3D rendering of the left hemisphere from one of the eight subjects. The surface represents the boundary between the white and gray matters, which were identified using the segmentation algorithm (see the Methods and Materials section). A parallel treatment of data from the rotating hemifield stimulus is shown in Figure 3a–c. The color also indicates the phase of the periodic response, which was proportional to the polar angle of the local visual field representation. The locations of several visual areas were identified by measuring the angular visual field representations.

Figure 3.

a–c: The top row shows a wedge map coded by color (blue areas are the boundaries of the dorsal V1, V2, V3; purple areas are the boundaries of the ventral V1, V2, V3) displayed on the original cortical surface (a), the inflated cortical surface (b), and the cut and flattened cortical surface (c). d,e: The bottom row shows retinotopy of eccentricity representation in the visual cortex by fMRI mapping for central (d) and peripheral (e) vision areas. Phase-encoded eccentricity maps were rendered on close-up views of the inflated left hemisphere (left) and right hemisphere (right).

Figure 4.

Results for presentation of contralateral and ipsilateral stimuli in both hemispheres of one participant. MT+ subdivision and localizer were activated by wide-field stimulation. a–f: The fMRI responses on an unfolded map, centered within the fundus of the occipital continuation of the ITS. The strong response depicted in a,b is evident throughout MT+. Colors correspond to correlation values above threshold (t > 4.0). c–f: Stimulus location is indicated by the labels in the left field of c,d and the right field of e,f. The distinct, anterior subregion (MST) responded to ipsilateral stimulation. When the stimulus is contralateral (d,e) to a particular hemisphere, an area encompassing both MT and MST is activated. When the stimulus is ipsilateral (c,f), only the MST area is activated.

Maps of the response in relation to an expanding ring on a medial view of the cortical surface are shown in Figure 3d,e, with the color indicating the eccentricity (distance from the center) that caused the activation. The hue of the color at each cortical surface point indicates the response phase, which is proportional to the eccentricity of the local visual field representation. A systematic increase in eccentricity (0–7.5°, 7.5–22.5°, 22.5–45°, 45–60°) is observed, moving anteriorly along the medial wall of the occipital cortex. As the expanding ring stimulus moved from the center to the periphery of the retina, the location of the responding areas varied from the posterior to anterior portions of the calcarine sulcus in what is referred to as the eccentricity dimension of the retinotopic map. The larger peripheral representation crossed to the fundus of the parieto-occipital sulcus (POS).

For each moving stimulus (random dots), area MT+ was defined and divided into MT and MST (Fig. 4). We found that different responses to the moving dots used in this wide-view visual presentation system can be used to subdivide the region of MT+. MT+ was identified (Fig. 4a,b) separately for each subject, based on a combination of anatomical and functional criteria. Specifically, a contiguous region was marked by hand to include voxels on the lateral surface of the occipital lobe, where the fMRI time series correlated strongly with the moving/stationary stimulus alternations (r > 0.25, chosen separately for each subject). Area MST was also defined separately for each subject to include a contiguous subregion of MT, distinct from retinotopically defined MT that responded strongly to peripheral, ipsilateral stimulation. Figure 4c,f shows the ipsilateral responses in the right hemisphere of one subject. Although ipsilateral responses were relatively weak compared with contralateral responses, a subregion of peripheral ipsilateral activity was clearly identifiable, marked by the cyan curve drawn on the unfolded map.

Quantitative Comparison of These Results With Those From Previous Studies

Table 1 contains measurements of the surface area of the visual field representations from 0 to 60°. Right and left hemispheres (left and right visual fields) are listed separately for each subject. The surface area of V1 shows an approximately two to threefold size variation (right hemisphere = 1916–2474 mm2, mean = 2223 mm2; left hemisphere = 1822–2595 mm2, mean = 2181 mm2) in 16 hemispheres (Table 1). This mean and range of variability are similar to those reported by Andrews et al in their anatomical study of 28 hemispheres (15).

Table 1. Visual Area Sizea
 Left hemisphereRight hemisphere
SubjectV1V2V3MT+V1V2V3MT+
  • a

    Surface area measurements for the 120° visual field representation of V1, V2, V3 and MT+ areas. Right and left hemispheres (left and right visual fields) are listed separately for each subject.

BY210220121775973222817211518867
DB237411541250971191617601470672
FZ222015501573466216016541490353
HO241021561895855221827541985819
KS182218431496689216420121695658
LS190116791380876202820041591817
NK259512821076572226715721445579
WH236717631480689247420071782522
Left mean222316791490761    
Right mean218119351622660    
Total mean220218071556711    
SD215370232123    

Figure 5 shows the surface area for 60° eccentricity stimuli and compares it with the previous research eccentricity stimuli (12°) and anatomical study (Fig. 5A). The mean V1 surface area is defined as 1470 mm2 from 0 to 12° visual field stimuli by fMRI research (7); the mean V1 area of this study is 2202 mm2 as determined by the wide-view (0–60°) stimuli presentation system, and the mean V1 surface area is defined as 2492 mm2 through anatomical analysis (Fig. 5B). The MT+ area of this study is 757 mm2, while the MT+ area reported by Huk et al is 562 mm2 (19).

Figure 5.

Comparison of our V1 and MT+ results with the previous research results. a: The V1 area of this study is 2202 mm2; the V1 area defined by fMRI is 1470 mm2 (Dougherty et al., 2003); the anatomical result for the V1 area is 2134 mm2 (15); the largest V1 area is 3102 mm2 (15). b: The MT+ area of this study is 757 mm2, while the MT area reported by Huk et al is 562 mm2 (19).

DISSUSSION

We developed a wide-view visual presentation system for fMRI using nonmagnetic optical fibers and a contact lens. This system can produce visual stimuli at a horizontal visual angle of 120° and a vertical angle of 120°. It not only solves the problems of stimuli that are presented to the central or peripheral visual fields due to the restricted space of the fMRI environment (1, 5), but also solves the problems of presenting stimuli that are not isotropic, as well as the high cost of visual presentation systems used for fMRI (12,19). We used a high-magnification contact lens to train the subject's eye to focus on the fiber center. As compared to previously developed existing visual systems, a contact lens can simplify the optical structure (20).

The wide-view visual stimulus presentation system can be used to define the retinotopic areas in the medial occipital lobe. Our results are in line with previous maps of occipital visual areas (1–3) and extend to larger eccentricities (60°). We found strong responses in the periphery visual field, and we were able to map the occipital retinotopic areas up to approximately 45–60° of eccentricity with monocular stimuli. Stimulation from center to far periphery field enables almost complete delineation of the retinotopic areas. In accordance with the histological data (14, 15), our results show that most peripheral representation in the retinotopic areas extends in depth through the cerebrum and to the deepest activation areas (Fig. 3). Among all eccentricity stimuli, the results are best for delineation of retinotopic areas, although the V2, V3 central part is missing, which is possibly due to the fovea in the retina [Fig. 3C] (21).

Wide-field retinotopic mapping revealed that the retinotopic organization and neighbor relations of human V6 closely resemble those reported for macaque V6 (12). Our research results confirm this finding by the wide-view visual presentation stimulus system (22). When the wide-field stimulus (60° eccentricity) was used, part of V6 (Fig. 3C) was activated. Previous magnetoencephalographic (21) and fMRI (23) studies using similar stimuli and approaches failed to reveal retinotopic organization along the POS, possibly due to the relatively small size of the stimuli they used; even their peripheral stimulation covered only the central 20° of the visual field, which resulted in the activation of only a small, lateral part of the area. The wide-field mapping stimuli also resulted in improved maps in the lateral occipital cortex and MT+. In particular, the polar-angle maps confirmed the presence of an anterior-facing upper-field representation in MT+, in line with previous reports (16). Previous works reported the existence of coarse retinotopic maps within the motion-responsive MT+ region (24–26). The use of wider-field stimuli in our study produced clearer results.

In conclusion, we developed a wide-view visual presentation system and used it quantitatively to investigate the properties of peripheral visual field representation, which provided data similar to the anatomical results obtained for V1 and MT+ areas of the human visual field (19). We plan to continue the study to improve the optical properties of the system and to try to build a more versatile setup that enables higher resolution stimulation in different types of head coils. With these developments, visual fMRI could become a clinical tool not only for probing the retinocortical pathway and preoperative mapping of the V1 but also for quantitative parametric studies of striate and extrastriate visual cortical functions.

Acknowledgements

We thank Dr. Kewei Chen and Dr. Takanori Kochiyama for their help in the experiment design. We aslo thank the subjects who participated in this study and the staff of the Osaka Neurosurgery Hospital for their assistance with experiment data collection. This study was financially supported by JSPS AA Science Platform Program and JSPS Grant-in-Aid for Scientific Research (B) (21404002).