IMPROVING THE SPEED AND ACCURACY OF THE P300 AUTOMATIC SPELLING SYSTEM THROUGH FACIAL RECOGNITION
Locked-in syndrome is a condition where an individual does not have control of their muscles, including facial muscles and vocal cords which control speech. Instead of oral speech, sign language, or writing, other methods must be used to achieve communication. Brain-computer interfaces relay the intentions of these individuals by using brainwave patterns such as event-related potentials (ERPs) that are collected using electroencephalogram electrodes. An ERP is a time-locked response with many features such as the P300 component. One way to elicit a strong ERP is using the “oddball” paradigm. When a stimulus is presented to the brain, it is classified as a target or non-target event. One of the two events categories is considered “rare” with a lower probability of occurrence than the other. When the “rare” event occurs, an ERP is evoked, with the amplitude of the P300 response being proportional to how small of a probability the event has of occurring. The rarer the stimuli, the stronger the P300 response will be. A P300 spelling system is a brain-computer interface that can allow the P300 component to be converted to text digitally by using signal processing and machine learning techniques. Using EEG measurements, the computer is trained to recognize features in a particular individual’s brainwaves. Once these features are identified, they are classified as target and non-target, allowing the computer to decide the desired outcome. However, current P300 spelling systems are still much slower than conventional communication and can be inaccurate. This research aims to improve the speed of the system without compromising accuracy by testing stimulus type and stimulus duration. Specifically, the standard paradigm will be altered to include a familiar face overlay with the intent of generating a stronger P300 component which can be more easily classified as a target or non-target stimulus. Different flash durations will also be tested in an attempt to improve the speed of spelling.
Albanese, Thomas. (July 2022). IMPROVING THE SPEED AND ACCURACY OF THE P300 AUTOMATIC SPELLING SYSTEM THROUGH FACIAL RECOGNITION (Master's Thesis, East Carolina University). Retrieved from the Scholarship. (http://hdl.handle.net/10342/11099.)
Albanese, Thomas. IMPROVING THE SPEED AND ACCURACY OF THE P300 AUTOMATIC SPELLING SYSTEM THROUGH FACIAL RECOGNITION. Master's Thesis. East Carolina University, July 2022. The Scholarship. http://hdl.handle.net/10342/11099. October 02, 2022.
Albanese, Thomas, “IMPROVING THE SPEED AND ACCURACY OF THE P300 AUTOMATIC SPELLING SYSTEM THROUGH FACIAL RECOGNITION” (Master's Thesis., East Carolina University, July 2022).
Albanese, Thomas. IMPROVING THE SPEED AND ACCURACY OF THE P300 AUTOMATIC SPELLING SYSTEM THROUGH FACIAL RECOGNITION [Master's Thesis]. Greenville, NC: East Carolina University; July 2022.
East Carolina University