Scholarlink

Motivation <p>Patient with Locked-In Syndrome (LIS) are unable to communicate through conventional means despite retaining full cognitive function. This condition severely impacts their autonomy and quality of life. Existing Brain-Computer Interface (BCI) solutions are often invasive, expensive, or require complex calibration and visual engagement, limiting their practical usability in clinical or home environments.</p> Objective <p>This study proposes <i>Brainlink</i>, a non-invasive EEG-based BCI system to provide a reliable means of binary communication (yes/no) for locked-in patients. The objective is to provide an accessible, safe, and adaptive communication system to empower these individuals to interact with their environment. It also serves as a valuable tool for caregivers and healthcare professionals to better understand and respond to the patient's intentions.</p> Methodology <p>The system uses the FlexCap EEG headset by Neuphony, configured to utilize 8 critical electrode channels (Fp1, Fp2, F3, F4, F7, F8, C3, C4), to capture the brain neural signals. These signals undergo preprocessing and are classified into three categories—left-sided, right-sided, and no action—using a deep learning model with spatial–temporal feature extraction. The system adapts responses using a binary tree-based question structure and allows caregivers to customize the question set based on patient needs.</p> Implementations and Results <p>The proposed approach is simulated using four models on the EEG Motor Movement/Imagery Dataset (EEGMMI). The LSTM + CNN + Oversampling model achieved the highest performance with an average accuracy of 91% and a best-case accuracy of 95%. A prototype questionnaire application further demonstrates the system’s ability to dynamically navigate questions based on user responses, with potential for real-world use in clinical settings.</p> Conclusion <p><i>Brainlink</i> offers a significant advancement in BCI-based communication for locked-in patients by addressing major limitations of existing solutions. It is non-invasive, cost-effective, and adaptable, making it a scalable and user-friendly solution that can meaningfully improve quality of life and caregiving effectiveness.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Brainlink: Voice for Patient with Locked-in Syndrome Using EEG Based BCI and Deep Learning

  • 期刊论文

  • 发表日期：

    2026年5月10日

查看全文
  • Adamya Singh,
  • Aryash Bajaj,
  • Abhyuday Singh,
  • Shikha Jain

摘要

Motivation

Patient with Locked-In Syndrome (LIS) are unable to communicate through conventional means despite retaining full cognitive function. This condition severely impacts their autonomy and quality of life. Existing Brain-Computer Interface (BCI) solutions are often invasive, expensive, or require complex calibration and visual engagement, limiting their practical usability in clinical or home environments.

Objective

This study proposes Brainlink, a non-invasive EEG-based BCI system to provide a reliable means of binary communication (yes/no) for locked-in patients. The objective is to provide an accessible, safe, and adaptive communication system to empower these individuals to interact with their environment. It also serves as a valuable tool for caregivers and healthcare professionals to better understand and respond to the patient's intentions.

Methodology

The system uses the FlexCap EEG headset by Neuphony, configured to utilize 8 critical electrode channels (Fp1, Fp2, F3, F4, F7, F8, C3, C4), to capture the brain neural signals. These signals undergo preprocessing and are classified into three categories—left-sided, right-sided, and no action—using a deep learning model with spatial–temporal feature extraction. The system adapts responses using a binary tree-based question structure and allows caregivers to customize the question set based on patient needs.

Implementations and Results

The proposed approach is simulated using four models on the EEG Motor Movement/Imagery Dataset (EEGMMI). The LSTM + CNN + Oversampling model achieved the highest performance with an average accuracy of 91% and a best-case accuracy of 95%. A prototype questionnaire application further demonstrates the system’s ability to dynamically navigate questions based on user responses, with potential for real-world use in clinical settings.

Conclusion

Brainlink offers a significant advancement in BCI-based communication for locked-in patients by addressing major limitations of existing solutions. It is non-invasive, cost-effective, and adaptable, making it a scalable and user-friendly solution that can meaningfully improve quality of life and caregiving effectiveness.