Objective Assessment of Beat Quality in Transcranial Doppler Measurement of Blood Flow Velocity in Cerebral Arteries

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Transcranial Doppler (TCD) ultrasonography is crucial for measuring cerebral blood flow velocity.
• Accurate clinical interpretation of TCD signals relies on identifying and excluding low-quality data points ('beats').
• Existing methods for TCD signal quality assessment can be labor-intensive and subjective.

Purpose of the Study:

• To develop and validate an automated algorithm for identifying and removing poor-quality beats from TCD signals.
• To enhance the reliability and efficiency of TCD data analysis.
• To improve the estimation of clinically significant TCD parameters.

Main Methods:

• An iterative outlier detection algorithm was developed using objective waveform features (e.g., Euclidean distance, cross-correlation, power ratios, beat length, diastolic variance).
• The algorithm was tested on over 15 hours of TCD data from 48 stroke patients and 34 controls.
• Performance was evaluated by comparing algorithm-derived TCD parameters against manual beat annotations.

Main Results:

• The algorithm demonstrated strong correlation with manual beat annotation, confirming successful recovery of clinically relevant features.
• Significant improvements were observed in estimating TCD parameters when using algorithm-accepted beats compared to all beats.
• The automated method proved effective in identifying reliable data segments.

Conclusions:

• The developed algorithm serves as a valuable tool for clinicians, enabling automated detection of reliable TCD data.
• This automated approach can serve as a pre-processing step to enhance data quality for machine learning-based diagnosis of pathologic beat waveforms.
• The findings suggest potential for improved diagnostic accuracy and efficiency in TCD analysis.