Detection of Monophasic Slow-wave Activation Phase Using Wavelet Decomposition

Area of Science:

• Physiology
• Bioengineering
• Signal Processing

Background:

• Gastric bio-electric slow-waves are crucial for stomach motility.
• Extracellular recordings provide physiological insights into slow-wave characteristics.
• Automated detection methods are needed for large datasets and long recordings.

Purpose of the Study:

• To develop and validate an automated algorithm for detecting the gastric slow-wave activation phase.
• To compare the performance of the novel Variable Threshold Wavelet (VTW) algorithm against an existing method.

Main Methods:

• The Variable Threshold Wavelet (VTW) algorithm was developed using wavelet decomposition to compute derivatives for activation phase detection.
• Performance was evaluated by comparing VTW against the Falling-Edge, Variable Threshold (FEVT) algorithm.
• Synthetic noise (ventilator and high-frequency) was added to in vivo slow-wave recordings for testing.

Main Results:

• The VTW algorithm demonstrated comparable performance to the FEVT algorithm in the presence of ventilator noise.
• In high-frequency noise conditions, VTW improved the area under the curve (A_roc) metric by 11.1% compared to FEVT.
• Key performance metrics included sensitivity, positive-predictive value, A_roc, and percentage improvement metric (PIM).

Conclusions:

• The VTW algorithm offers a reliable and accurate method for automated detection of gastric slow-wave activation phases.
• VTW shows enhanced performance in noisy environments, particularly with high-frequency interference.
• This algorithm can be applied to analyze both normal and abnormal gastric slow-wave recordings.