A Survey on Medical Image Analysis in Capsule Endoscopy

Area of Science:

• Medical Imaging
• Computer Vision
• Gastroenterology

Background:

• Capsule endoscopy (CE) offers a non-invasive alternative to traditional endoscopy but generates lengthy videos, complicating abnormality detection.
• Gastroenterologists face challenges with CE videos, including length, required concentration, and subjective interpretation of abnormalities.
• Automated abnormality detection systems are crucial for improving diagnostic accuracy and efficiency in capsule endoscopy.

Purpose of the Study:

• To review and analyze image analysis techniques for automated abnormality detection in capsule endoscopy.
• To provide a comparative analysis of various approaches, their performance, strengths, and limitations.
• To identify challenges and future directions in capsule endoscopy image analysis.

Main Methods:

• A systematic review of research papers from IEEE, Scopus, and Science Direct databases was conducted.
• Search criteria included "capsule endoscopy," "engineering," and "journal papers."
• 62 selected publications focusing on image analysis were rigorously reviewed.

Main Results:

• The review covers key aspects of medical image analysis for CE: video summarization, image enhancement, segmentation, abnormality detection, and compression.
• A comparative analysis of different techniques, experimental setups, performance metrics, strengths, and limitations is presented.
• Various computer-aided detection methods for capsule endoscopy abnormalities were evaluated.

Conclusions:

• Current image analysis techniques for capsule endoscopy have not fully addressed all challenges.
• The complexity of the gastrointestinal tract and the lack of comprehensive datasets are significant limitations.
• Further research is needed to overcome existing hurdles in automated capsule endoscopy analysis.