A novel method for automatically locating the pylorus in the wireless capsule endoscopy

Area of Science:

• Medical Imaging
• Gastroenterology
• Computer Vision

Background:

• Wireless capsule endoscopy (WCE) is crucial for non-invasively examining the digestive tract.
• The stomach and small intestine are high-risk areas for infections, necessitating accurate organ localization.
• Current WCE image analysis requires robust methods for identifying specific anatomical structures.

Purpose of the Study:

• To propose a novel, automated method for pylorus localization in WCE images.
• To improve the accuracy and efficiency of identifying the boundary between the stomach and small intestine.
• To provide a reliable tool for clinicians in diagnosing gastrointestinal diseases.

Main Methods:

• A two-level localization approach: rough-level color change detection and refined-level feature extraction.
• Utilizing an improved Weber Local Descriptor (WLD) for enhanced feature extraction robustness in grayscale images.
• Employing a K-nearest neighbor (KNN) classifier for image segmentation around approximate pylorus positions.

Main Results:

• The algorithm successfully identified three probable pylorus locations, matching clinician markings.
• The refined-level analysis demonstrated improved robustness in feature extraction compared to original WLD methods.
• The proposed method effectively segments WCE images for pylorus localization.

Conclusions:

• The developed automated pylorus localization method is effective for WCE image analysis.
• This technique offers a significant advancement in the diagnostic capabilities of WCE.
• Accurate pylorus localization aids in the early detection and management of gastrointestinal conditions.