COVID-19 ground-glass opacity segmentation based on fuzzy c-means clustering and improved random walk algorithm

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Artificial Intelligence in Medicine

Background:

• Accurate segmentation of ground-glass opacity (GGO) is crucial for diagnosing COVID-19.
• Existing GGO segmentation methods struggle with adhesive GGO connected to anatomical structures.

Purpose of the Study:

• To propose an accurate GGO segmentation method for COVID-19 detection.
• To address limitations of current methods in segmenting complex GGO structures.

Main Methods:

• A novel approach combining fuzzy c-means (FCM) clustering and an improved random walk algorithm.
• Integration of a Markov random field (MRF) with adaptive spatial information into the FCM model.
• Adaptive weighting and an adaptive snowfall model for noise reduction and edge preservation.

Main Results:

• The proposed method demonstrates improved accuracy in segmenting GGO, even in challenging cases.
• Experimental results show superior performance compared to traditional and state-of-the-art segmentation techniques.
• The method effectively suppresses noise while preserving critical edge details.

Conclusions:

• The developed method offers a reliable tool for auxiliary diagnosis of COVID-19.
• This approach can significantly enhance the efficiency of medical professionals in diagnosing COVID-19.