Cerebral visual impairment: Documenting visual effects and simplifying visual life
With investigations into CVI in the early phases, collaboration across the Atlantic is crucial
Eye tracking was used to quantify gaze behavior, such as the extent of the visual search area explored and the number of fixations made. Image credit: ©NDABCREATIVITY – stock.adobe.com
Cerebral (or cortical) visual impairment (CVI), the most common cause of visual impairment in paediatric patients in developed countries,1 is the umbrella term for brain-based visual disorders that damage and/or cause maldevelopment of the retrochiasmal visual processing areas in patients without a major ocular disease.
At first glance, these patients in many cases can appear to have normal or nearly normal visual acuity levels. However, many of these patients have perceptual visual deficits, a cardinal feature of CVI, which are not typically tested during a standard ophthalmology office evaluation.
Until recently, clinicians could only guess at what affected patients were seeing with any accuracy. Lotfi Merabet, OD, PhD, from the Laboratory for Visual Neuroplasticity in the Department of Ophthalmology at Massachusetts Eye and Ear at Harvard Medical School in Boston, explains that anecdotal evidence provided a strong clue that suggested that children with CVI could have difficulties identifying objects presented as abstractions or cartoons.
These are “representations of images that depart from naturalistic forms and shapes as well as lack key associative information provided by colour. In contrast, individuals with CVI are more likely to correctly identify these same objects when presented as coloured photographs,” Merabet explains.
CVI collaboration leverages the best of both worlds
Merabet described findings from a recent study2 conducted in collaboration with local teachers of individuals with visual impairment, researchers from the Translational Vision Lab in the Department of Psychology at Northeastern University in Boston, Massachusetts, and investigators from the Unit of Child Neurology and Psychiatry at Azienda Socio Sanitaria Territoriale degli Spedali Civili of Brescia in Italy. This team’s members have done pioneering work to unravel the intricacies of CVI in children in the United States and Italy. The two countries employ different approaches investigating CVI, and the collaboration combines the strong points of both continents, according to Merabet.
The Europeans’ work is based on the medical model of CVI: Children born with a brain injury are evaluated and given a diagnosis early at the hospital, are followed through the national health care system and remain under focus within that system. In contrast, in the United States, visual concerns in a child with CVI are often first noticed by parents and teachers of those with visual impairment. At the same time, because universal diagnostic criteria are lacking, many children with CVI can fall through the cracks of the system.
An interesting observation is that in the United Kingdom, the health care system has taken things a step further. In that country, the focus is on identifying and caring for mothers whose children may develop this neurodevelopmental disorder, effectively trying to cut the disease off at the knees. The US system tends to be more reactionary in that once CVI is identified, the resources are then focused on affected patients, which helps explain the socioeconomic bias of health care in the United States. “Because investigations into CVI are in the early phases, the collaboration across the Atlantic [Ocean] is crucial,” Merabet comments.
The pathway to CVI discovery
The buildup to the US/Italian study was based on an observation provided by Matthew Tietjen, a study coauthor and teacher of individuals with visual impairment who noticed a difference in how children with CVI perceived images with their classroom materials. Many of the illustrations were abstractions and cartoons, and he noticed that the children would often misname the objects.
“This is interesting from a scientific standpoint because of the need to understand why this happens as well as…[its] important educational repercussions. In many cases, children with CVI have to work harder than typically sighted children to get through school material,” Merabet states.
In one anecdote, Tietjen showed a child a picture of a cartoon elephant in frontal view. The child identified the elephant as a Sony PlayStation remote controller, presumably mistaking the ears of the elephant for the handles of the controller and the elephant’s eyes for the dark-coloured joysticks.
“As the pictures get further and further from reality, it is harder for children with CVI to understand what they are looking at,” Merabet says. “It has a lot to do with how information is taken into the brain, how it is analysed and processed, stored, and assigned meaning. The more abstract an image, the more tenuous the object identification.”
This observation by Tietjen resulted in his developing a more systematic and standardised approach to test his observation, with the goal of identifying the drivers of the misidentifications expressed by the children. Based on an object-naming assessment developed by Tietjen, he and Merabet assembled a series of images matched for size, perspective, complexity and familiarity called the Two-Dimensional Image Study.2
The investigators chose 12 images of common animate and inanimate objects, each represented from five possible image categories. An example taken from the article is the five categories for a cat (ie, colour photo, realistic color drawing, black-and-white realistic outline drawing, colour abstract drawing and black-and-white abstract outline drawing of a cat). These representations would disentangle colour from form cues and use the colour photo as the benchmark for identification.
The representations of the 12 objects in the five categories being evaluated were shown to children on a computer, who pressed the space bar when they named the objects. During this exercise, the investigators used an eye tracker to record where the children were looking from the moment the image appeared to when they identified (or misidentified) an object. Eye tracking was used to quantify gaze behavior, such as the extent of the visual search area explored and the number of fixations made.
The US (n = 50) and Italian (n = 50) children who participated in the study were not statistically different in terms of age, sex and disease manifestations. However, Merabet suspects that the US children tended to be from a higher socioeconomic bracket than those in Europe, where the health care system tends to be more accessible.
Figure. Heat maps showing the distribution of fixation points in control compared to CVI participants. Note (1) the greater distribution and extent of fixation points in CVI compared to controls and (2) the greater distribution of fixation points viewing the abstract outline drawing compared to the colour photo image of the same object. (Images courtesy of Lotfi Merabet et al)
Manley et al2 reported that compared with controls, the children with CVI “showed significantly lower success rates and longer reaction times when identifying objects. In the CVI group, the success rate improved moving from abstract black-and-white images to colour photographs, suggesting that object form (as defined by outlines and contours) and colour are important cues for correct identification.”
Furthermore, the results obtained using eye tracking with the controls and patients in the study were markedly different in that the controls exhibited tighter gaze patterns, ie, one fixation to the body and one fixation to the head of the object. In contrast, in patients with CVI, the area of the gaze pattern was much larger and with a greater number of fixations, indicating that the children seemed to explore a much larger area (Figure), particularly if the object was misidentified. Finally, the distribution of the eye gaze patterns in the CVI group was less aligned with the high saliency features of the image compared with controls,” he reports.
He also notes the importance of having benchmarks to know which features in the images are important to the patients to identify the pictures correctly. The investigators used Graph-Based Visual Saliency (GBVS), a mathematical analysis that looks at an image and identifies the features in the image that can be considered the most salient from a bottom-up standpoint, such as colour, luminance, and edges.
The investigators used the information (the predictive ability) provided by GBVS and combined it mathematically with the individual patient’s gaze pattern, yielding a mathematical comparison between a standard and a response that informs the investigators about how well the two outcomes agree. The study findings concluded that the results have important implications in helping to understand the complex profile of visual perceptual difficulties associated with CVI. “The takeaway is that the patients with CVI fared worse than controls in all five outcomes,” he reports.
Data analysis showed that the controls scored 100% on image identification. The children with CVI scored lower than controls in all outcomes. “As they moved from abstract to outline to colour photo, they improved slightly and identified the images faster, the receiver operating curve trended upward, and the number of fixations decreased,” he says.
The visual search area did not have a clear pattern as a function of the type of image. However, the patients searched a much larger area compared with the controls. For children with CVI, when they erred in their identification of an image, they took more time to do so.
Considerations
Merabet points out that the exercises described underscore the importance of listening to the observations of the community, such as teachers and parents who often have tremendous insight into the visual difficulties of their students and children. The experience also shows that leveraging the international collaboration provided a larger sample size of patients, allowing for more statistically robust findings. With larger study populations, there is potential to explore questions regarding differences in factors such as geography, disease manifestations, age, race, and socioeconomic status.
The implications of these findings affect choices of educational materials used in the classroom. For children with CVI, abstractions and outlines may represent a greater cognitive burden and lost learning time. “For these children to thrive, we must design a world for them that considers their visual needs,” Merabet says.
A pragmatic approach to address CVI is to simplify the visual world for these children by using images with high relevance (ie, those close to the real world) and minimising clutter. Giving them more time to process images and checking to see whether they are identifying things properly are vital. “There are nuances in terms of education that are extremely important. The most important takeaway is that you cannot take for granted that they are following a lesson and understanding the content just because their peers are following,” Merabet says.
Finally, Merabet advises that research in CVI should move forward as a community endeavour and strive to answer questions that can have an impact on the lives of these individuals. Without this effort, he believes, his team would not have achieved the level of research that they did without this kind of collaboration.
The investigators are currently working on a study in which patients with CVI are presented with a real-world image of a lamp and then instructed to find it in a picture of a room. In a second version of the task, a text cue is used (the word “lamp”) and the person has to read it and find the object in a picture of a room. This analysis is ongoing, but early results suggest the success rate in CVI is lower and the reaction time is longer to carry out this task. However, one interesting finding is that using a text cue slowed the patients more compared with a visual cue.3
He pointed out that for children with CVI, their verbal IQ score, which provides an index of an individual’s overall verbal intellectual abilities, was related to performance. The higher the verbal IQ score, the greater their success rate and lower their reaction time in identifying objects. This leads to the intriguing possibility that developmental outcomes such as verbal IQ score may be important indices of how well an individual can perceive and interact with their complex visual surroundings. The ultimate goal is to understand how individuals with CVI interact with their visual world and to improve accessibility to educational materials for all students regardless of visual abilities.
References
1. Solebo AL, Teoh L, Rahi J. Epidemiology of blindness in children. Arch Dis Child. 2017;102(9):853-857. doi:10.1136/archdischild-2016-310532
2. Manley CE, Walter K, Micheletti S, et al. Object identification in cerebral visual impairment characterized by gaze behavior and image saliency analysis. Brain Dev. 2023;S0387-7604(23)00085-2. doi:10.1016/j.braindev.2023.05.001
3. Manley CE, Walter K, Bex PJ, Merabet LB. Visual search patterns in cerebral visual impairment (CVI) are driven by saliency cues when exploring naturalistic scenes. Abstract presented at: Vision Sciences Society Meeting; 19-24 May, 2023; St Pete Beach, FL.
Abstract 4941.
Lotfi Merabet, OD, PhD | E: lotfi_merabet@meei.harvard.edu
Dr Merabet is from the Laboratory for Visual Neuroplasticity in the Department of Ophthalmology at Massachusetts Eye and Ear at Harvard Medical School in Boston and serves on the Board of Directors of the Perkins School for the Blind in Massachusetts. He has no financial interest in this subject matter.
