Artificial intelligence-based detection of pharyngeal cancer using convolutional neural networks

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Pharyngeal cancer often presents at advanced stages, leading to poor prognosis.
• Early detection of superficial pharyngeal cancer is challenging despite advancements like narrow-band imaging (NBI).
• Artificial intelligence (AI) shows promise in enhancing medical diagnostic capabilities.

Purpose of the Study:

• To evaluate the diagnostic performance of an AI-based system for detecting pharyngeal cancer in endoscopic images.
• To assess the AI system's ability to identify both superficial and advanced pharyngeal cancers.

Main Methods:

• A retrospective collection of 5403 training images from 247 pharyngeal cancer cases (202 superficial, 45 advanced).
• Development of a deep learning system using convolutional neural networks (CNNs).
• Validation using 1912 images from 75 patients (40 with pharyngeal cancer, 35 without).

Main Results:

• The AI system achieved 100% detection of pharyngeal cancer lesions (40/40) in validation images.
• Sensitivity for detecting pharyngeal cancer was 85.6% with NBI and 70.1% with white light imaging.
• The AI system analyzed 1912 images in just 28 seconds.

Conclusions:

• The AI-based diagnostic system demonstrates high sensitivity in detecting pharyngeal cancer.
• This technology has the potential to facilitate earlier diagnosis, improving patient prognosis and quality of life.
• The AI system offers a rapid and effective tool for identifying pharyngeal cancers.