Review of electrooculography-based human-computer interaction: recent technologies, challenges and future trends

Linkai Tao, Hao Huang, Chen Chen, Loe Feijs, Jun HuConnected 2023 · DOI

Abstract: Electrooculography-based Human-Computer Interaction (EOG-HCI) is an emerging field. Research in this domain aims to capture eye movement patterns by measuring the corneal-retinal potential difference. This enables translating eye movements into commands, facilitating human-computer interaction through eye movements. This paper reviews articles published from 2002 to 2022 in the EOG-HCI domain, aiming to provide a comprehensive analysis of the current developments and challenges in this field. It includes a detailed and systematic analysis of EOG signal electrode arrangement, hardware design for EOG signal acquisition, commonly used features, and algorithms. Representative studies in each section are presented to help readers quickly grasp the common technologies in this field. Furthermore, the paper emphasizes the analysis of interaction design within the EOG-HCI domain, categorizing different interaction task types and modalities to provide insights into prevalent interaction research. The focus of current research in the field is revealed by examining commonly used evaluation metrics. Lastly, a user-centered EOG-HCI research model is proposed to visually present the current research status in the EOG-HCI field from the perspective of users. Additionally, we highlight the challenges and opportunities in this field.

Introduction: Addressing the Challenges of Non-Invasive HCI

In the field of non-invasive brain-computer interfaces (BCIs) and assistive technologies, identifying reliable "triggers" for user commands remains a significant hurdle, particularly for individuals with severe motor impairments such as Amyotrophic Lateral Sclerosis (ALS). While camera-based eye tracking is a common solution, it is often hampered by environmental factors like lighting changes, the inability to track closed eyes, and high computational requirements. Similarly, Electroencephalography (EEG) is a popular modality but is limited by low signal voltage (< 100 μV) and a high susceptibility to noise.

Electrooculography (EOG) offers a robust alternative by measuring the corneal-retinal potential difference, which provides a relatively strong (> 1 mV) and stable signal. Despite twenty years of research, the EOG-HCI field has remained fragmented. This review synthesizes literature from 2002 to 2022 to map the current landscape, categorize hardware and algorithmic approaches, and identify the bottlenecks preventing mainstream adoption.

Methodology: A Systematic Review of Two Decades of Research

The authors conducted a systematic review, screening 299 papers and selecting 35 high-quality journal articles that presented a complete EOG-HCI loop. These studies were analyzed across several technical and design dimensions:

  • Hardware and Electrode Configuration: Setups were categorized into Standard (clinical), Task-Oriented (minimal electrodes for specific functions), and Wearable-Oriented (integrated into everyday items like glasses or headbands).
  • Signal Processing and Algorithms: The review details preprocessing steps (e.g., bandpass filtering, wavelet transforms) and classification methods, ranging from simple threshold detection and waveform matching to traditional machine learning (SVM) and modern neural networks.
  • Interaction Design Framework: A novel analytical framework was introduced, categorizing applications into Control-type (e.g., wheelchair navigation) and Information input-type (e.g., virtual keyboards).
  • Evaluation Metrics: The authors proposed the Interaction Benefit score (C) to quantify the trade-off between an application's Degree of Interaction Freedom (number of commands) and its Interaction Efficiency (eye movements per command).
  • User-Centered Research Model: A five-level hierarchical model was proposed to guide future research: Usefulness $\rightarrow$ Availability $\rightarrow$ Usability $\rightarrow$ Intelligence $\rightarrow$ Enjoyment.

A critical finding regarding data was the severe lack of open, standardized EOG datasets tailored for HCI. Most reviewed studies relied on small-scale, proprietary data collected in controlled laboratory settings, which has hindered the application of Deep Learning in this domain.

Key Insights: The Current State of EOG-HCI

The analysis revealed several critical trends and limitations within the current research landscape:

  1. The Algorithm Bottleneck: Most EOG-HCI systems rely on basic threshold detection. While computationally efficient, this method is brittle, requiring frequent recalibration and struggling with complex command structures.
  2. Underutilization of Deep Learning: Despite its dominance in other biosignal fields, Deep Learning was utilized in only four of the reviewed studies. The field remains largely in a "feasibility exploration" phase.
  3. Visual Overload and Cognitive Tax: Many designs require users to focus on flashing GUI elements to trigger commands. This "visual-for-visual" interaction can be cognitively taxing and potentially hazardous in real-world environments where vision is needed for environmental awareness.
  4. Engineering vs. Human Factors: Research heavily favors system performance metrics (e.g., accuracy) over usability. User-related metrics, such as the System Usability Scale (SUS), appeared half as frequently as engineering-centric metrics in the literature.

Discussion and Future Directions

To move EOG-HCI from the laboratory to practical, mainstream application, the authors identify several necessary shifts:

  • Development of Open Multimodal Datasets: Large-scale, open-access EOG databases including physiological and psychological metadata are essential for training robust, generalizable models.
  • Adoption of Advanced Architectures: Shifting from thresholding to architectures like Graph Neural Networks (GNNs) could enable the capture of high-dimensional, sample-independent features, reducing the need for user-specific calibration.
  • Context-Aware Interaction: Integrating EOG with computer vision (e.g., outward-facing cameras) would allow systems to understand the intent behind eye movements based on the user's environment.
  • Standardized UX Evaluation: Incorporating standardized usability scales (SUS, UMUX) is vital to ensure that systems are not only technically functional but also practical and comfortable for the end-user.

Conclusion

EOG-HCI represents a promising frontier for assistive technology, offering a high-signal, non-invasive method for interaction. However, the field must transition from simple feasibility studies toward robust, user-centered applications powered by deep learning and context-aware design. By addressing the current lack of standardized data and prioritizing human factors, EOG-HCI can achieve its potential as a reliable interface for users with motor impairments.

Summary updated

Abstract

Electrooculography-based Human-Computer Interaction (EOG-HCI) is an emerging field. Research in this domain aims to capture eye movement patterns by measuring the corneal-retinal potential difference. This enables translating eye movements into commands, facilitating human-computer interaction through eye movements. This paper reviews articles published from 2002 to 2022 in the EOG-HCI domain, aiming to provide a comprehensive analysis of the current developments and challenges in this field. It includes a detailed and systematic analysis of EOG signal electrode arrangement, hardware design for EOG signal acquisition, commonly used features, and algorithms. Representative studies in each section are presented to help readers quickly grasp the common technologies in this field. Furthermore, the paper emphasizes the analysis of interaction design within the EOG-HCI domain, categorizing different interaction task types and modalities to provide insights into prevalent interaction research. The focus of current research in the field is revealed by examining commonly used evaluation metrics. Lastly, a user-centered EOG-HCI research model is proposed to visually present the current research status in the EOG-HCI field from the perspective of users. Additionally, we highlight the challenges and opportunities in this field.
BibTeX
@inproceedings{tao2023review,
  author = {Linkai Tao and Hao Huang and Chen Chen and Loe Feijs and Jun Hu and Wei Chen},
  title = {{Review of electrooculography-based human-computer interaction: recent technologies, challenges and future trends}},
  booktitle = {Connected},
  year = {2023},
  doi = {10.20517/chatmed.2023.05},
}
Figure