Generalized Feature Extraction for Wrist Pulse Analysis: From 1-D Time Series to 2-D Matrix

Area of Science:

• Biomedical Engineering
• Medical Informatics
• Traditional Chinese Medicine

Background:

• Traditional Chinese pulse diagnosis relies on subjective experience, leading to diagnostic inconsistencies.
• Objective analysis of wrist pulse waveforms is crucial for scientific study.
• Existing pulse feature extraction methods often treat pulse signals as simple 1-D time series, neglecting valuable information.

Purpose of the Study:

• To present a generalized method for wrist pulse feature extraction, extending from 1-D to 2-D matrix representation.
• To validate the proposed 2-D method through pattern classification using actual pulse records.
• To compare the effectiveness of the 2-D generalized features against conventional 1-D features in diagnostic applications.

Main Methods:

• Developed a generalized method for pulse feature extraction, transforming 1-D time series data into a 2-D matrix.
• Utilized pattern recognition theories for analyzing the extracted 2-D pulse features.
• Validated the method through pattern classification, specifically applied to diabetes diagnosis using real pulse data.

Main Results:

• The generalized 2-D matrix feature effectively extracts both periodic and non-periodic information from wrist pulse signals.
• Quantitative and qualitative results demonstrate the superiority of the 2-D features over 1-D features in diabetes diagnosis.
• The proposed method proves practical and effective for objective wrist pulse analysis.

Conclusions:

• The generalized 2-D matrix feature extraction method offers a more comprehensive approach to wrist pulse analysis.
• This advanced method enhances diagnostic accuracy by capturing richer information compared to traditional 1-D techniques.
• The findings support the application of advanced signal processing and pattern recognition in objective Traditional Chinese Medicine diagnostics.