Bleeding detection in wireless capsule endoscopy videos - Color versus texture features.

Area of Science:

• Medical Imaging
• Gastroenterology
• Computer Vision

Background:

• Wireless capsule endoscopy (WCE) generates extensive video data, making manual analysis of gastrointestinal (GI) tract abnormalities time-consuming.
• Accurate and efficient detection of GI bleeding from WCE is crucial for timely diagnosis and treatment.

Purpose of the Study:

• To develop and evaluate an automated technique for detecting bleeding in WCE frames.
• To improve the efficiency and accuracy of GI bleeding diagnosis using WCE data.

Main Methods:

• Proposed an automated bleeding detection technique utilizing color and texture features.
• Employed machine learning-based classification for distinguishing bleeding from non-bleeding frames.
• Implemented pixel-level segmentation for precise localization of bleeding areas.

Main Results:

• The automated method achieved high detection accuracy, comparable to state-of-the-art approaches.
• Effectively distinguished between bleeding and non-bleeding frames using combined color and texture features.
• Demonstrated the system's capability for pixel-level segmentation of bleeding regions.

Conclusions:

• The proposed automated WCE bleeding detection system offers an efficient and flexible solution for clinical applications.
• The combination of color, texture features, and machine learning provides robust performance in identifying GI bleeding.
• This approach significantly reduces the burden of manual frame-by-frame analysis for clinicians.