QC Speller: User Interface Design of a Hands-Free Touch-Free Speller with Brain Electroencephalogram Sensory Rhythm
Tulika Nama, Debasis Samanta · 2025 · ACM Transactions on Accessible Computing · doi:10.1145/3705733
Summary
This paper presents QC Speller, a novel brain-computer interface (BCI) text entry system designed for people with severe motor impairments who cannot use traditional keyboards, mice, or touchscreens. The system uses electroencephalography (EEG) to detect motor imagery signals—the neural patterns generated when users imagine moving their left or right hand—and translates these into keyboard control commands. The key innovation is a quarter-circle user interface that requires only two control commands (left-hand and right-hand motor imagery) to access 36 target symbols, compared to three or four commands needed by prior BCI spellers. The interface arranges letters by frequency using Huffman coding principles, placing commonly used letters like E, T, A, and O in positions requiring fewer selections. This optimization significantly improves text entry speed compared to alphabetically arranged interfaces. The system architecture consists of two integrated subsystems: a front-end user interface built with PyQt5 and a back-end BCI processing pipeline using an OpenBCI Cyton 8-channel board with ThinkPulse dry electrodes. Signal processing includes common spatial pattern (CSP) filtering and a sparse representation-based classifier (SRC) that handles the variability of EEG signals across users and sessions. The researchers developed four prototype iterations, progressively refining the interface based on usability findings before arriving at the final Prototype-IV design.
Key findings
The QC Speller was evaluated with 10 participants (ages 24-34) recruited from the Indian Institute of Cerebral Palsy and NIEPMD, representing diverse motor impairments including cerebral palsy (hemiplegia, diplegia, monoplegia), muscular dystrophy, and spinal cord injury. None had prior BCI experience. Key results: - Mean text entry rate of 5.20 characters per minute (error-free) and 6.04 CPM including errors - Mean entry accuracy of 98.04% across all participants - Correction efficiency of 0.69, meaning users could fix most errors with minimal effort - Highest individual performance: 6.14 CPM error-free with 99.31% accuracy (Subject-B, limb-girdle muscular dystrophy) - Offline classification accuracy of 89.62% for the motor imagery classifier Comparative analysis showed QC Speller significantly outperformed two existing MI-based BCI spellers (Virtual Keyboard and Hex-o-spell) on the same participants. The mean error-free entry rate was 4.59 CPM for QC Speller versus 1.58 CPM for Virtual Keyboard and 2.16 CPM for Hex-o-spell. The introduction of six cursor navigation keys for phrase-level and word-level error correction proved particularly effective, with 83% of users preferring these keys over character-level backspacing.
Relevance
This research advances hands-free, touch-free text entry for people with severe motor impairments who have no functional limb movement. The practical implications are significant: users only need to imagine left or right hand movements—no actual physical movement required—making this viable for individuals with quadriplegia, advanced ALS, or severe cerebral palsy. The study's participant diversity (cerebral palsy, muscular dystrophy, spinal cord injury) strengthens the ecological validity for real-world deployment. The use of consumer-grade OpenBCI hardware rather than expensive medical-grade EEG systems suggests potential for home use. However, limitations include the controlled laboratory environment, short text phrases, and the 20-minute setup time. Future work should address extended use fatigue and integration with word prediction. The error correction mechanism is a significant contribution—designing BCI interfaces that gracefully handle the inherent noisiness of EEG classification is essential for practical adoption.
Tags: brain-computer interface · motor imagery · EEG · text entry · motor impairment · speller · assistive technology · cerebral palsy