Vision-Based Assistive Technologies for People with Cerebral Visual Impairment: A Review and Focus Study
Bhanuka Gamage, Leona Holloway, Nicola McDowell, Thanh-Toan Do, Nicholas Price, Arthur Lowery, Kim Marriott · 2024 · ASSETS '24: Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility · doi:10.1145/3663548.3675637
Summary
This paper reveals a significant research gap in assistive technology for people with cerebral visual impairment (CVI) through a scoping review and focus studies. CVI is caused by damage to the brain's visual processing centres rather than the eyes themselves, making it fundamentally different from ocular visual impairments like macular degeneration, glaucoma, or cataracts. While the eyes may function normally, the brain cannot properly process visual information, resulting in challenges with visual complexity (difficulty perceiving objects in cluttered scenes), simultanagnosia (inability to perceive multiple objects at once), impaired motion perception (dorsal stream dysfunction), difficulty with facial recognition, reduced visual field awareness, and problems with visual guidance of movement. CVI is now the leading cause of visual impairment in children in developed countries and is increasingly recognised in adults, yet the scoping review of 110 vision-based assistive technology (VBAT) papers found that virtually none addressed CVI specifically — the vast majority targeted ocular low vision conditions. The authors conducted three focus studies with seven participants with CVI (four adults with acquired CVI and three parents of children with CVI) to explore their daily challenges, current coping strategies, and opportunities for assistive technology. Sessions covered navigation, object recognition, reading, social interaction, and digital device use. The paper compares the assistive needs of people with CVI against those with ocular low vision, demonstrating that many VBAT approaches designed for ocular conditions (magnification, contrast enhancement, edge detection) may be ineffective or even counterproductive for CVI because the underlying problem is not optical but neurological.
Key findings
The scoping review categorised existing VBAT into five functional areas: scene understanding and object recognition, text reading, navigation and wayfinding, social interaction (face recognition, emotion detection), and image enhancement (magnification, contrast, edge highlighting). For ocular low vision, image enhancement techniques like magnification and contrast boosting are effective because the visual processing system is intact — the problem is getting sufficient signal to the brain. For CVI, however, the processing itself is impaired: magnifying a cluttered scene does not help someone with simultanagnosia who cannot perceive multiple objects at once; increasing contrast may actually worsen visual overload. Focus study participants described their primary challenges as visual clutter and complexity rather than acuity — crowded supermarket shelves, busy streets, and complex digital interfaces were overwhelming because their brains could not parse multiple visual elements simultaneously. Scene simplification — reducing visual complexity by highlighting relevant objects and suppressing background clutter — emerged as the most promising VBAT direction for CVI. Participants valued AR overlays that could isolate target objects from their surroundings, reduce visual noise in real-world scenes, and provide audio descriptions when visual processing failed entirely. Motion perception difficulties meant that moving objects (approaching vehicles, people walking) were particularly hazardous, suggesting a need for motion-alert systems. Participants also noted that their CVI fluctuated with fatigue, stress, and sensory overload — they could process visual information reasonably well when rested but experienced dramatic declines when tired, creating a need for adaptive systems that respond to changing ability levels throughout the day.
Relevance
This paper makes a compelling case that CVI represents a major blind spot in assistive technology research. As the leading cause of childhood visual impairment in developed countries, CVI affects a growing population that is systematically underserved by existing VBAT designed around ocular models of vision loss. The fundamental distinction between optical and neurological visual impairment has profound design implications: the standard assistive technology toolkit for low vision (magnification, high contrast, large text) may be insufficient or harmful for CVI. Instead, CVI-appropriate interventions centre on reducing complexity, simplifying scenes, isolating relevant information, and adapting to fluctuating processing capacity — principles that also benefit people with cognitive accessibility needs more broadly. For accessibility practitioners, the paper highlights the danger of treating "visual impairment" as a monolithic category: the needs of someone with CVI are radically different from someone with macular degeneration, even though both may have similar acuity measurements. The fluctuation finding connects to ability-based design — systems for CVI users must be able to sense and adapt to changing visual processing capacity in real time. The call to action for the HCI and AT research community is clear: CVI needs dedicated attention, not assumptions that solutions for ocular impairments will transfer.
Tags: cerebral visual impairment · CVI · vision-based assistive technology · computer vision · augmented reality · image enhancement · machine learning · scoping review · visual processing