Performance of an Automated Detection Algorithm to Assess Objective Pulsatile Tinnitus.

Area of Science:

• Otolaryngology
• Biomedical Engineering
• Medical Acoustics

Background:

• Pulsatile tinnitus (PT) is often diagnosed subjectively, leading to potential inconsistencies.
• Objective diagnostic methods for PT are needed to differentiate it from other forms of tinnitus.

Purpose of the Study:

• To develop and evaluate an automated algorithm for objective detection of pulsatile tinnitus.
• To assess the algorithm's performance against subjective clinical judgment.

Main Methods:

• Sensitive microphones recorded ear canal sounds and heart rate in 36 PT patients.
• A novel algorithm calculated a pulsatility index based on sound-heart rate coherence across 6 octave bands.
• Algorithm performance was compared to the consensus of three blinded observers.

Main Results:

• The automated detection algorithm demonstrated good agreement with blinded observers (ROC AUC 0.83).
• Substantial interobserver reliability (Fleiss's κ=0.64) was observed among the human evaluators.
• The algorithm effectively identified patterns indicative of pulsatile tinnitus.

Conclusions:

• The developed algorithm offers a reliable, objective alternative to subjective assessments for pulsatile tinnitus detection.
• This tool can aid clinicians in differentiating objective pulsatile tinnitus from subjective forms.
• Automated analysis of in-canal sound measurements provides a valuable objective measure in PT diagnosis.