Automatic segmentation of blood vessels from dynamic MRI datasets

Area of Science:

• Medical imaging
• Radiology
• Biomedical engineering

Background:

• Rheumatoid arthritis (RA) diagnosis and monitoring require accurate visualization of inflammation.
• Dynamic contrast-enhanced MRI (DCE-MRI) is used to assess joint inflammation in RA.
• Blood vessels can obscure inflammation signals in DCE-MRI, necessitating their exclusion for precise evaluation.

Purpose of the Study:

• To develop and validate a blood vessel segmentation approach for hand joint DCE-MRI in RA patients.
• To enable accurate visualization of inflammatory activation events.
• To facilitate objective evaluation of inflammation severity.

Main Methods:

• Statistical modeling incorporating physiological properties of contrast agent uptake.
• Application of a Markov random field probabilistic framework.
• Utilizing principal component analysis for feature extraction.

Main Results:

• The developed algorithm successfully segmented blood vessels from hand joint DCE-MRI datasets.
• The method demonstrated promising results in excluding vasculature for improved inflammation visualization.
• Tested on 60 temporal slices, indicating robustness.

Conclusions:

• The proposed blood vessel segmentation technique is effective for DCE-MRI in RA patients.
• Accurate vessel exclusion enhances the objective assessment of joint inflammation.
• This approach holds potential for improving RA diagnosis and treatment monitoring.