Alzheimer's Disease Classification With a Cascade Neural Network

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Alzheimer's Disease (AD) diagnosis is challenging, especially for early detection of mild cognitive impairment (MCI).
• Current methods using only gait or electroencephalogram (EEG) data have limitations, including poor detection of MCI/AD differences or time-consuming data collection and loss of crucial spatial-temporal information.
• Simultaneous analysis of gait and EEG data offers a promising avenue for more accurate and efficient AD classification.

Purpose of the Study:

• To develop a faster and more accurate method for classifying Alzheimer's Disease (AD) by integrating gait and electroencephalogram (EEG) data.
• To improve the detection of mild cognitive impairment (MCI) and differentiate between healthy controls (HC), MCI, and AD stages.
• To leverage the complementary strengths of gait and EEG data for a comprehensive diagnostic approach.

Main Methods:

• A two-step cascade neural network was proposed, utilizing both gait and EEG data.
• The first step employed attention-based spatial-temporal graph convolutional networks on Kinect-captured skeleton sequences (gait) to distinguish HC from patients.
• The second step used spatial-temporal convolutional networks on EEG data to classify patients into MCI or AD categories, preserving spatial and temporal information.

Main Results:

• The proposed method achieved a significantly higher accuracy of 91.07% in the three-way classification (HC, MCI, AD) compared to existing methods (68.18%).
• The study identified the lower body and right upper limb as critical for early AD diagnosis based on gait analysis.
• The integrated approach effectively captured spatial and temporal information from EEG, overcoming limitations of frequency-domain analysis.

Conclusions:

• The novel cascade neural network effectively integrates gait and EEG data for accurate and efficient Alzheimer's Disease classification.
• This method shows significant potential for early detection of MCI and differentiation of AD stages.
• The findings regarding the importance of specific body parts in gait analysis offer valuable insights for clinical research and early AD diagnosis.