Research on Algorithm for Feature Extraction of Laryngoscope Image Distribution and Texture Fusion

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Otolaryngology

Background:

• Increasing medical data burdens physicians, necessitating advanced diagnostic tools.
• Computer-assisted diagnosis for laryngopharyngeal reflux (LPR) using laryngoscope images requires further development.

Purpose of the Study:

• To develop an innovative feature extraction and classification method for laryngoscope imaging data.
• To improve the accuracy and efficiency of diagnosing LPR using computer-assisted methods.
• To alleviate diagnostic workload for physicians and provide practical clinical value.

Main Methods:

• Utilized laryngoscope images from Jilin University Second Hospital.
• Developed an integrated feature extraction method combining local binary patterns (texture) and gray histogram (distribution) features.
• Compared the performance of five classic classification algorithms on the extracted features.

Main Results:

• The proposed feature extraction method combined with the random forest algorithm achieved 96.61% accuracy in laryngoscope image classification.
• The developed algorithm demonstrated excellent classification performance.
• The method showed low sample size requirements, indicating high efficiency and practicality.

Conclusions:

• The integrated feature extraction and random forest classification method offers a highly accurate and efficient approach for LPR diagnosis.
• This computer-assisted method has significant practical value in clinical settings.
• The study highlights the potential of advanced image analysis in reducing physician diagnostic burden.