Peter Bex, Ph.D.

Profile

peter_bex.jpg

Senior Scientist, Schepens Eye Research Institute

Associate Professor, Department of Ophthalmology, Harvard Medical School

 

 

 

 

A recent addition to the Schepens faculty, Peter Bex received his PhD in Vision Science from Cardiff University, UK, in 1994. His thesis project was an applied study of perceptual failures experienced by pilots reading dynamic information on computer-generated cockpit displays. This was followed by post-doctoral research positions at McGill University in Montreal and at the University of Rochester, NY. Peter studied optics, used laser systems to present images directly on the retina, and began to examine visual function in the natural environment. To read more about Dr. Bex, please click here.

Please click here to listen to Dr. Bex's recent interview on the Royal National Institute for the Blind's Insight Radio, broadcast in the U.K.  (mp3 file)

Education

1992 – 1994 Ph.D. CASE studentship. “Temporal Aliasing in Computer Graphics Animation”. Department of Psychology, University of Cardiff, UK. Supported by British Aerospace plc and by the Science and Engineering Research Council.

Contact Information

617-912-2528
FAX: 617-912-0111
This e-mail address is being protected from spambots. You need JavaScript enabled to view it.

Bex Laboratory on You Tube

Click image below to view YouTube video coverage of the Bex Laboratory's Head-mounted eye tracking application for driving.

This video represents the Master's Thesis Project of Nicolas Schneider, in co-operation with the Institute of Neuro- and Bioinformatics, University of Luebeck, and the Schepens Eye Research Institute, harvard Medical School. Click for more information about the ITU Gaze tracker...

bex_lablg.jpg

Research Story

A key objective of current behavioral vision research concerns the early detection, diagnosis, and monitoring of visual impairment.

A recent addition to the Schepens faculty, Peter Bex, received his PhD in Vision Science from Cardiff University, UK, in 1994. His thesis project was an applied study of perceptual failures experienced by pilots reading dynamic information on computer-generated cockpit displays. This was followed by post-doctoral research positions at McGill University in Montreal and at the University of Rochester, NY. Peter studied optics, used laser systems to present images directly on the retina, and began to examine visual function in the natural environment. Under natural conditions, existing computational models of visual processing are frustrated because of the complex and dynamic content of real scenes compared with laboratory or clinical stimuli. Peter then returned to the UK to take up a faculty position at the Institute of Ophthalmology in London in 2000, where he concentrated on translational research between basic and clinical vision science. Our understanding of visual processing is largely based on studies of the central vision of healthy, young subjects; Peter began to update our understanding of visual processing to deal with the effects of ageing and eye disease. At Schepens, he is continuing this work and applying it to clinical populations, including those suffering from glaucoma, age-related macular degeneration, and amblyopia.

 

bex_1.jpg

 
Figure 1. Your blind spot. Close your right eye and look at the airplane on the right or close your left eye and look at the airplane on the left. When you move closer to the page (around 6 inches or so), the other plane will fall within your blind spot and disappear. Just as we are unaware of this blind area of our vision, people are often unaware of blind areas caused by eye disease until the loss has become catastrophic. 

A fundamental challenge confronting clinical vision scientists concerns the remarkable plasticity and redundancy of the human visual system. For example, we are usually unaware that objects falling within the blind spot (where the optic nerve leaves the eye) are not visible, as shown in Figure 1. Likewise, blind spots caused by neurodegenerative disease or retinal insult may go undetected until significant vision loss has occurred, by which time it may be too late for the most effective treatments to be offered. A key objective of current behavioral vision research concerns the early detection, diagnosis, and monitoring of visual impairment.This endeavor is also a cornerstone for the development and assessment of new treatments and neuro-protective approaches, where sensitive measurement of visual function and dysfunction forms an essential part of evaluating treatment outcomes. 

Detection of Glaucoma

Glaucoma causes steady loss of peripheral vision and is the second leading cause of blindness in the developed world. Increases in life expectancy mean that around 80 million people worldwide are expected to have glaucoma by 2020. Early treatment for glaucoma can significantly reduce the progression of the disease but any lost vision is currently irreversible. It is thus essential to diagnose the disease as early as possible; however, most screening techniques for glaucoma work only once significant vision loss has already occurred.
Researchers are, therefore, working on new behavioral techniques that ask people to detect or discriminate carefully specified images. An understanding of the functional roles of different classes of retinal cells is an essential part of this process. For example, some retinal cells are involved in color vision, while others are insensitive to color but contribute to the perception of motion. Likewise, some cells have rapid response times, while others respond more slowly. We now know that some classes of retinal cell are more vulnerable to the early stages of glaucoma than others and we can combine this knowledge with an understanding of their stimulus selectivity and response properties to refine stimuli and tasks that isolate only the most vulnerable cells. These cells should show a sensitivity loss before other classes of cell and we can use this sensitivity loss for early diagnosis and to evaluate new and emerging neuro-protective interventions.

Visual Function in Age-Related Macular Degeneration


Age-related macular degeneration (AMD) causes blindness in the high-resolution area of central vision. Approximately 12 million people suffer from AMD and this figure is set to rise as our population ages. Treatment of macular disease with conventional ophthalmic techniques is of limited benefit in the majority of cases, forcing people to depend on their poor-resolution peripheral vision and severely impairing essential tasks such as mobility, face recognition and reading.

bex_2.jpg

Figure 2. Crowding. If you fixate on the cross, the letter E is the same size (magnification), contrast and distance on the left and right. However, the adjacent letters on the right make it much harder to identify the middle letter. This effect is called crowding and it is crowding rather than acuity that limits vision across the majority of the visual field. 

The structure of the peripheral retina of people with AMD can be relatively intact, yet it supports greatly impoverished vision. This is true even in normally sighted people. To experience this for yourself, try reading words a couple of lines below where you are currently looking or recognizing the expressions of people if you look at the top of their heads. This phenomenon, illustrated in Figure 2, is known as crowding and it is a key factor limiting visual function in people with AMD. Vision scientists are trying to understand how visual processing and eye-movement behavior differ between central and peripheral visual fields. We are trying to learn the functional organization of the visual system that is responsible for these differences, so that we can develop new assistive devices that circumvent these problems. We are also beginning to understand some of the mechanisms of perceptual learning and neural plasticity that underlie the improvements in visual sensitivity seen in training programs. Techniques that minimize the effects of crowding, along with training that helps people learn to use their residual vision more effectively, will drive visual rehabilitation programs. Furthermore, our understanding of perceptual learning processes will be a critical stage for training people to use any new prosthetic and regenerated retinal implants that may be developed in the future.  

Amblyopia


Abnormal binocular vision in childhood can lead to the development of amblyopia in the absence of any observable ocular pathology. Amblyopia is commonly known as ‘lazy eye’ and is the leading cause of visual impairment in childhood, affecting approximately 3% of the population. The visual experience of amblyopia varies, but is often described as distorted rather than blurred, as illustrated in Figure 3b.

bex_3.jpg

With no evident ocular pathology, treatments for amblyopia depend on perceptual learning techniques. The main treatments for amblyopia temporarily impair vision in the better eye with eye patches or drops. These treatments force the child to use her amblyopic eye, which would otherwise be suppressed by the better eye. Treatment is effective in around 75% of juvenile cases, but in fewer than half of adult cases. Relatively little is known about how treatment works and why is sometimes doesn’t, whether full stereo vision is restored, and levels of recidivism. Vision scientists are currently trying to understand the normal and abnormal development of binocular vision and to develop more-structured training programs to combine with patching or eye drops. Recent data from animal studies suggest that correcting the de-correlation between the images in each eye (for example, when an image resembles 3a with one eye, but 3b or 3c with the other) is critical for reversing amblyopia. It is likely that new, more-widely effective treatments for amblyopia will involve a correction for binocular de-correlation.

Laboratory Members

Luis Lesmes, PhD

Research Associate

Michael Dorr, PhD

Postdoctoral Fellow

William J. Harrison, PhD

Postdoctoral Fellow

MiYoung Kwon, PhD

Postdoctoral Fellow

Guido Maiello

Graduate Student

Alexandra Miller

Undergraduate Student

Christopher Taylor, PhD

Postdoctoral Fellow

Thang Long To, PhD

Postdoctoral Fellow

Emily Wiecek

PhD Candidate

Laboratory Alumni

Lotte-Guri Bogfjellmo

John Cass, PhD

Michael Crossland, PhD

Helle Falkenberg, PhD

John Greenwood, PhD

David Kane, BSc

Isabelle Mareschal, PhD

Lee McIlreavy, PhD

Nicolas Schneider, BSc

Anita Simmers, PhD

Jonas Vibell, PhD

Catherine Vishton

Susan Wardle, PhD

Richard Watson, PhD

Thomas Wallis, PhD

Research Projects

My research employs behavioral and imaging techniques to study the human visual system in normal and abnormal development, in normal ageing and in neurodegenerative diseases. Particular emphasis is placed on the use of natural stimuli and tasks to examine the underlying mechanisms of visual processing.

Equivalent Noise Analysis in Glaucoma

Glaucoma is the second leading cause of blindness in the developed world and causes steady loss of peripheral visual field. Increases in life expectancy mean that around 80 million people worldwide will have glaucoma by 2020. While visual loss caused by glaucoma is currently irreversible, clinical intervention can significantly reduce its progression and is most effective when the disease is caught in its early stages, when catastrophic field loss can be averted. However, existing screening techniques can only detect glaucoma once significant visual impairment from gross loss of retinal ganglion cells (RGC) has occurred. The most sensitive screening for glaucoma must therefore identify early signs of RGC dysfunction before irreversible cell death . We are currently developing screening tests, based on visual sensitivity to moving images that should be able to detect early signs of glaucoma before existing techniques can. We use a technique known as equivalent noise analysis that estimates the level of internal noise in the visual system by measuring sensitivity to the direction of motion as more and more external noise is added to a stimulus, see Figure 1 for illustration. The rate at which external noise impairs performance can be used to compare the level of dysfunctional and non-functional sensors in the normal and glaucomatous visual system.

 a b c
4a.gif 4b.gif 4c.gif

 

Figure 1 Equivalent noise analysis can be used to estimate the level of internal noise in the visual system and the sampling efficiency of visual sensors. It measures how quickly performance, in this case a direction of movement task, declines as (b) more and (c) more external noise is added to the movie. We attribute elevated levels of internal noise to dysfunctional motion sensors and a loss in efficiency to non-functional motion sensors

 Publications

  1. Falkenberg, H.K. and P.J. Bex, Sources of Motion Sensitivity Loss in Glaucoma. Invest Ophthalmol Vis Sci, 2007, 48.p. 2913-2921Bex, P.J. and H.K. Falkenberg, Resolution of complex motion detectors in the central and peripheral visual field. Journal of the Optical Society of America A, 2006. 23 (7): p. 1598-607.

  2. Dakin, S.C., I. Mareschal, and P.J. Bex, Local and global limitations on direction integration assessed using equivalent noise analysis. Vision Research, 2005. 45 (24): p. 3027-49.

Age-Related Macular Degeneration

Approximately 12 million people suffer central vision loss caused by Age-related Macular Degeneration (AMD - www.eyesight.org), a figure that is set to rise as our population ages. Treatment of macular disease with conventional ophthalmic techniques is of limited benefit in the majority of cases, forcing people to depend on their poor resolution peripheral vision and severely impairing essential tasks such as mobility, face recognition and reading. We are developing new ways to present information, images and text that may maximise residual vision and independence in people with AMD. In principle it should be possible to use magnification to compensate for the loss of resolution in the peripheral visual field. However, even magnified images can be completely masked by nearby features, an effect known as crowding , and this is particularly problematic in the peripheral visual field.   Figure 2 illustrates how difficult it can be to disambiguate features in the periphery, especially under crowded conditions. We are studying the visual processes that cause crowding in an effort to develop novel ways to present information that are not vulnerable to its effects.


bex5.gif

Figure 2 Crowding. When you fixate the red squares, details in the top image are difficult to disambiguate in this complex scene. Identical image patches are much more visible when nearby crowding details are masked, without changing the size or contrast of the patches. This shows that it is usually crowding rather than acuity that limits vision in the peripheral visual field.

 

 

Publications

  1. Bex, P.J. and S.C. Dakin, Spatial interference among moving targets.  Vision Research, 2005. 45 (11): p. 1385-98.

  2. Bex, P.J., S.C. Dakin, and A.J. Simmers, The shape and size of crowding for moving targets. Vision Research, 2003. 43 (27): p. 2895-904.

  3. Falkenberg, H.K., G.S. Rubin, and P.J. Bex, Acuity, crowding, reading and fixation stability. Vision Research, 2007 in press

  4. Falkenberg, H.K. and P.J. Bex, The Impact of Central and Peripheral Visual Field Loss on Eye Movements and Mobility While Walking Invest Ophthalmol Vis Sci 2005; 46: E-Abstract 4608.

Rapid Perimetry

Diseases such as glaucoma or diabetic retinopathy and other causes of retinal insult, such as laser injury, cause blind spots that often go undetected. This is because the scotomas are often in different locations in the two eyes and because perceptual processes fill-in the scotomas and render them invisible, as they do in normal vision at the blind spot where the optic nerve leaves the eye. The occurrence of filling-in means that simple charts, known as campimetry, cannot be used reliably to help people locate their own blind spots. Instead, point-wise testing across the visual field, called micro-perimetry, is required, but this can be painstakingly slow. We have developed a new behavioral technique that can identify blind spots quickly and easily.

 a  6c.gif    6a.gif
6b.gif  c

Figure 3 Afterimage Perimetry. a) A blank patch embedded in dynamic noise (like a detuned TV set) eventually disappears and is ‘filled in' with the background pattern. b) If after a few minutes the noise is replaced with a blank field, a twinkling after-image is seen in the position occupied by the blank patch, but not elsewhere. Filling-in also occurs at the locations of visual field loss caused by eye disease and we have found that real scotomas also give rise to a twinkling afterimage. c) Patients can trace around the apparent locations of any afterimages on a touch-sensitivie screen and there is good agreement between the locations of the afterimages and scotomas measured by conventional microperimetry.

Figure 3 a) simulates a blind spot in a dynamic noise pattern and is often referred to as an artificial scotoma . After a short while, the blank patch disappears and the whole screen appears to contain noise. This effect can be easily demonstrated by sticking a small piece of paper in the centre of a detuned TV set and fixating the edge of the screen, so the square is in peripheral visual field. The paper gradually appears to be replaced by the TV noise. When the noise is switched off and replaced by a blank screen, paradoxically, the areas that did contain noise now appear blank, while the areas that were blank now appear to contain dynamic noise. We have found that this twinkling afterimage occurs for real scotomas caused by eye disease as well as for artificial scotomas. We have developed this observation into a simple test for undetected blind spots. Patients view a dynamic noise image alternated with a blank screen for several ten-second cycles. During the noise periods, the patients passively watch the screen and are free to move their eyes. During the blank periods they are asked to fixate a steady cross and trace around any anomalous areas on a touch-sensitive screen. The afterimages collected in this way in a few minutes closely correspond to those collected with microperimetry over 30 minutes or more. This test can be run at home on a TV or on a computer so that anyone can test themselves and self-refer at the first signs of a problem and secure the most effective early clinical intervention.

Publication

  1. Bex, P. J., Crossland, M. and Dakin, S. C. Afterimage Campimetry a New Technique for the Rapid Detection of Scotomas. Invest Ophthalmol Vis Sci 2007: E-Abstract 2348.

 

 

Amblyopia

Abnormal binocular vision in infancy can lead to the development of amblyopia, commonly known as ‘lazy eye', in which vision in the affected eye can be severely impaired. Amblyopia is the leading cause of visual impairment in childhood.

7a.gif
7b.gif

 

Figure 4 Visual distortions can cause a loss in visual acuity – the distorted letters on the right are harder to identify even at the same size and contrast. If only one eye suffers distortions, this can disrupt binocular vision and induce competition or rivalry between the eyes

Treatment for amblyopia usually involves patching or blurring vision in the unaffected eye and this intervention is effective in around 75% of juvenile cases, but less than half of adult cases. We have previously studied the perception of form and motion in amblyopia and found that visual distortions, like those illustrated in Figure 4, rather than blur are a key feature of amblyopia. We are developing new computer-based treatment and assessment techniques for amblyopia. We aim to monitor the effects of treatment on the sound eye and to restore binocular vision in juvenile and adult amblyopes who are not responsive to conventional treatment.

Publications

  1. Simmers, A.J. and P.J. Bex, The representation of global spatial structure in amblyopia.  Vision Research, 2004. 44 (5): p. 523-33.

  2. Hess, R.F., J.S. Pointer, A. Simmers, and P. Bex, Border distinctness in amblyopia. Vision Research, 2003. 43 (21): p. 2255-64.

  3. Simmers, A.J., P.J. Bex, and R.F. Hess, Perceived blur in amblyopia. Investigative Ophthalmology and Visual Science, 2003. 44 (3): p. 1395-400.

  4. Hess, R.F., R. Demanins, and P.J. Bex, A reduced motion aftereffect in strabismic amblyopia. Vision Research, 1997. 37 (10): p. 1303-11

 

Natural Scenes

A great deal of sensory research has attended to the problem of how we detect and encode elementary components of visual images (such as sine gratings) that are experimentally presented in isolation and under conditions that render them barely visible (e.g. at contrast detection threshold). However, relatively little is known about how these components are integrated for the perception of natural images, which contain many visual elements at a range of contrasts. We have been studying the statistics of static and dynamic natural images and relating the properties that characterize natural images to the sensitivity of the human visual system. In many studies, we have found that the visual system is optimized to process natural scenes.

Publications

  1. Dakin, S.C., I. Mareschal, and P.J. Bex, An oblique effect for local motion: psychophysics and natural movie statistics. Journal of Vision, 2005. 5 (10): p. 878-87.

  2. Bex, P.J., S.C. Dakin, and I. Mareschal, Critical band masking in optic flow. Network: Computation in Neural Systems, special issueS Sensory Coding in the Natural Environment 2005. 16 (2-3): p. 261-84.

  3. Dakin, S.C. and P.J. Bex, Natural image statistics mediate brightness 'filling in'. Proceedings of the Royal Society of London B: Biological Sciences, 2003. 270 (1531): p. 2341-8.

  4. Fiser, J., P.J. Bex, and W. Makous, Contrast conservation in human vision. Vision Research, 2003. 43 (25): p. 2637-48.

  5. Bex, P.J. and W. Makous, Spatial frequency, phase, and the contrast of natural images. Journal of the Optical Society of America A, 2002. 19 (6): p. 1096-106.

  6. Brady, N., P.J. Bex, and R.E. Fredericksen, Independent coding across spatial scales in moving fractal images. Vision Research, 1997. 37 (14): p. 1873-83.

  7. Bex, P.J., A.J. Simmers, and S.C. Dakin, Grouping local directional signals into moving contours. Vision Research, 2003. 43 (20): p. 2141-53.

  8. Mareschal, I., S.C. Dakin, and P.J. Bex, Dynamic properties of orientation discrimination assessed by using classification images. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103 (13): p. 5131-6.

  9. Bex, P.J., N. Brady, R.E. Fredericksen, and R.F. Hess, Energetic motion detection. Nature, 1995. 378 (6558): p. 670-2.