Scholarlink

<p>The human brain continuously transmits neural signals that coordinate voluntary movements such as walking or grasping objects. However, in Parkinson’s disease (PD), this communication is disrupted. Although neural signals reach their intended targets, the body fails to respond appropriately, resulting in movement impairments and coordination difficulties key symptoms of PD. Early detection is essential for managing symptoms before they significantly affect daily life. Traditional diagnostic approaches are often insufficient, as motor symptoms typically become evident only after substantial neuronal degeneration. Electroencephalography (EEG) has emerged as a non-invasive technique capable of detectin neurological changes associated with PD before overt motor symptoms develop. This study examines the impact of different EEG electrode configurations and sensory conditions on PD classification. A deep Convolutional Exponential Linear Unit Network (ConvELU-Net) is employed to analyze EEG signals. The Short-Time Fourier Transform (STFT) is applied to convert one-dimensional EEG signals into two-dimensional spectrograms, using a 1-second Hamming window with 50% overlap and decibel scaling normalization. Five experimental configurations are evaluated, ranging from full-scalp coverage (64 electrodes) to more localized setups, including left and right hemisphere-focused configurations (28 electrodes each). Experimental results indicate that the Left Hemisphere (28 electrodes) configuration achieves the highest classification accuracy of 98.40% under the Eyes Closed (EC) condition. Notably, using a reduced set of electrodes enhances signal clarity and computational efficiency, with region-specific configurations often outperforming full-scalp setups. These findings highlight the importance of targeted brain region analysis and sensory condition optimization for improving PD classification accuracy and facilitating early diagnosis.</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

Early detection of parkinson’s disease through electrode configurations and sensory states

  • 期刊论文

  • 发表日期：

    2025年8月9日

查看全文
  • Fatma Salah,
  • Amira Echtioui,
  • Yassine Ben Ayed

摘要

The human brain continuously transmits neural signals that coordinate voluntary movements such as walking or grasping objects. However, in Parkinson’s disease (PD), this communication is disrupted. Although neural signals reach their intended targets, the body fails to respond appropriately, resulting in movement impairments and coordination difficulties key symptoms of PD. Early detection is essential for managing symptoms before they significantly affect daily life. Traditional diagnostic approaches are often insufficient, as motor symptoms typically become evident only after substantial neuronal degeneration. Electroencephalography (EEG) has emerged as a non-invasive technique capable of detectin neurological changes associated with PD before overt motor symptoms develop. This study examines the impact of different EEG electrode configurations and sensory conditions on PD classification. A deep Convolutional Exponential Linear Unit Network (ConvELU-Net) is employed to analyze EEG signals. The Short-Time Fourier Transform (STFT) is applied to convert one-dimensional EEG signals into two-dimensional spectrograms, using a 1-second Hamming window with 50% overlap and decibel scaling normalization. Five experimental configurations are evaluated, ranging from full-scalp coverage (64 electrodes) to more localized setups, including left and right hemisphere-focused configurations (28 electrodes each). Experimental results indicate that the Left Hemisphere (28 electrodes) configuration achieves the highest classification accuracy of 98.40% under the Eyes Closed (EC) condition. Notably, using a reduced set of electrodes enhances signal clarity and computational efficiency, with region-specific configurations often outperforming full-scalp setups. These findings highlight the importance of targeted brain region analysis and sensory condition optimization for improving PD classification accuracy and facilitating early diagnosis.