Overseeing eyesight: Decoding visual stimuli from neural signals
Overseeing eyesight: Decoding visual stimuli from neural signals lead image
Vision is a complex process, with multiple steps and pathways to convert light into electrical signals in the brain. If one part in this system is damaged, artificial electrical stimulation can mimic stages in this process.
However, reproducing these electrical signals is challenging. Typically, studies using electrical stimulation rely on participants describing what they “see,” which suffers from subjectivity and inefficiency. Romeni et al. examined how electroencephalography (EEG) could be used to provide a more objective window into the brain’s electrical pathways.
EEG measures electrical signals in the brain in near real time. Though non-invasive, it is limited by poor spatial resolution and low signal-to-noise ratio, which makes reliable decoding difficult. Before EEG can oversee electrical stimulation in eyesight, it is essential to understand its responses.
“We placed healthy subjects in front of a large screen covering most of their visual field,” said Romeni. “We then flashed a series of rectangular bright shapes of various sizes and locations on a dark background, while recording EEG signals from 128 channels placed on the scalp. These stimuli can be imagined as the building blocks for simple visual scenes.”
The team used statistical analysis and machine learning techniques to identify features of the stimulus in the EEG responses and reduce the uncertainty in the feature identification process.
“Our work demonstrates that visual stimuli, which are compatible with the perceptions evoked by electrical stimulation, can be decoded from non-invasive EEG correlates,” said Romeni. “This capability could potentially be used for the automatic optimization of visual system stimulation in blind patients in the future.”
Source: “Decoding electroencephalographic responses to visual stimuli compatible with electrical stimulation,” by Simone Romeni, Laura Toni, Fiorenzo Artoni, and Silvestro Micera, APL Bioengineering (2024). The article can be accessed at https://doi.org/10.1063/5.0195680 .
