Iranian Journal of Biomedical Engineering (IJBME)
نوع مقاله : مقاله کامل پژوهشی
نویسندگان
1 گروه بیوالکتریک، دانشکده مهندسی پزشکی، دانشگاه صنعتی همدان، همدان، ایران
2 گروه مهندسی پزشکی دانشگاه صنعتی همدان
3 استادیار، گروه بیوالکتریک، دانشکده مهندسی پزشکی، دانشگاه صنعتی همدان، همدان، ایران
کلیدواژهها
موضوعات
عنوان مقاله English
نویسندگان English
Brain-Computer Interface (BCI) offers a communication pathway for individuals with severe disabilities, enabling them to spell words using brain activity. This study proposes a combined BCI system using Rapid Serial Visual Presentation (RSVP) and Steady-State Visual Evoked Potential (SSVEP) to enhance spelling accuracy and speed. The system organizes 36 characters into 3 groups of 12, each further divided into 4 subgroups of 3 characters. Each subgroup is arranged around a flashing square. SSVEP assigns unique frequencies to identify target groups, while triadic RSVP identifies subgroups. The target character is determined using single-frequency SSVEP.
Power Spectral Density Analysis (PSDA) detects SSVEP frequencies, and wavelet transform with Support Vector Machine (SVM) identifies P300 components in RSVP. Canonical Correlation Analysis (CCA) extracts features for a nonlinear SVM classifier in single-frequency SSVEP. The system was tested on 7 healthy participants, achieving 91.2% accuracy and an Information Transfer Rate (ITR) of 21.5 bits/min. Reducing SSVEP stimulation to 4 seconds maintained accuracy at 90.5% and increased ITR to 25.37 bits/min. Using Persian letters and more characters improved ITR while maintaining accuracy and reducing spelling time compared to similar studies.
کلیدواژهها English
