A subspace decomposition approach toward recognizing valid pulsatile signals.

Area of Science:

• Biomedical Engineering
• Neurology
• Signal Processing

Background:

• Automatic analysis of intracranial pressure (ICP) pulses aids in predicting neurological and cerebrovascular variations.
• Existing frameworks extract ICP pulse morphology but require improved methods for recognizing valid pulses.

Purpose of the Study:

• To enhance an existing ICP pulse analysis framework by incorporating Singular Value Decomposition (SVD).
• To replace the correlation-based approach for recognizing valid ICP pulses with a more accurate and efficient SVD technique.

Main Methods:

• Implemented SVD to replace the correlation-based method for identifying valid ICP pulses.
• Validated the SVD technique on a large dataset of 700 hours of ICP recordings from 67 neurosurgical patients.

Main Results:

• The SVD method achieved a 61.96% reduction in false positive rates while maintaining a 99.08% true positive rate.
• Computational time was reduced by 91.14% compared to the correlation-based method.
• Demonstrated significant improvements in both accuracy and processing speed.

Conclusions:

• The proposed SVD-based method offers superior accuracy and efficiency for recognizing valid ICP pulses.
• This technique has the potential for broader application in analyzing other pulsatile signals beyond ICP.
• The advancements support better management of neurological disorders through improved ICP signal analysis.