Video summarization based tele-endoscopy: a service to efficiently manage visual data generated during wireless...

Area of Science:

• Medical Imaging
• Gastroenterology
• Computer Science

Background:

• Wireless capsule endoscopy (WCE) offers advantages over traditional methods but generates vast amounts of data.
• Limited memory, energy, and communication capabilities pose challenges for WCE data management.
• Efficiently sharing and analyzing WCE video data is crucial for timely diagnosis.

Purpose of the Study:

• To develop a video summarization-based tele-endoscopy service for efficient WCE data handling.
• To estimate and extract semantically relevant video frames for gastroenterologists.
• To reduce data transmission and storage burdens while preserving diagnostic information.

Main Methods:

• Image moments, curvature, and multi-scale contrast were computed for each frame.
• A saliency map was generated by fusing these computed features.
• Keyframes were selected based on their diagnostic relevance using the saliency map.

Main Results:

• The proposed service effectively identifies and selects diagnostically relevant keyframes from WCE videos.
• Significant reductions in data transmission costs and bandwidth usage were achieved.
• Storage costs and physician browsing time were also reduced.

Conclusions:

• The developed tele-endoscopy service facilitates efficient WCE data collection and browsing.
• Keyframe selection based on diagnostic relevance ensures that important information is not lost.
• This approach enhances the utility of WCE in remote diagnostic support.