A semi-automatic method for peak and valley detection in free-breathing respiratory waveforms

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Commercial software often fails to accurately identify respiratory peaks in respiration-correlated computed tomography (RCCT).
• Precise respiratory phase detection is crucial for accurate RCCT image reconstruction and analysis.
• Variations in patient breathing patterns pose challenges for automated detection algorithms.

Purpose of the Study:

• To develop and validate a semi-automatic method for robust peak and valley detection in respiratory waveforms.
• To improve the accuracy and efficiency of respiratory phase determination for RCCT applications.
• To address the limitations of existing commercial software in handling free-breathing patient data.

Main Methods:

• A semi-automatic algorithm was developed to detect peaks and valleys in respiratory waveforms.
• Breath cycles were identified by intersecting a moving average curve with waveform branches.
• Peaks and valleys were defined as extrema between alternating inspiration and expiration intercepts.
• Automatic corrections and manual interventions were incorporated for enhanced accuracy.

Main Results:

• The method achieved an average automatic detection rate of 99% for peaks and valleys across 20 patients.
• Detection of an average of 307 peaks and valleys per patient was completed in just 2.8 seconds.
• The algorithm demonstrated robustness, effectively handling respiratory waveforms with significant variations.

Conclusions:

• The developed semi-automatic method provides accurate and efficient respiratory peak and valley detection.
• This technique offers a significant improvement over existing commercial software for RCCT.
• The robustness of the method supports its application in clinical settings with diverse patient breathing patterns.