Interactive iterative relative fuzzy connectedness lung segmentation on thoracic 4D dynamic MR images

Area of Science:

• Medical Imaging
• Radiology
• Computational Anatomy

Background:

• Accurate lung delineation in dynamic 4D thoracic MRI is crucial for analyzing pediatric respiratory diseases like thoracic insufficiency syndrome (TIS).
• Challenges include extreme thoracic malformations, poor image quality, abnormal dynamics, and patient cooperation issues in TIS.
• Existing methods like manual segmentation and registration-based approaches have limitations in efficiency and accuracy.

Purpose of the Study:

• To introduce and evaluate an interactive fuzzy connectedness approach for lung delineation in dynamic 4D thoracic MRI of pediatric TIS patients.
• To address the limitations of manual and automated segmentation methods in this challenging clinical context.

Main Methods:

• An interactive iterative relative fuzzy connectedness (IRFC) approach was developed, utilizing manually placed seeds propagated slice-by-slice.
• The method allows for efficient user control, with segmentation refinement based on user acceptance.
• Evaluation involved dynamic MRI from 5 pediatric TIS patients (39 3D volumes) and comparison with more automated IRFC strategies.

Main Results:

• The proposed interactive IRFC method achieved a high true positive volume fraction of 0.91 and a low false positive volume fraction of 0.03.
• The Hausdorff boundary distance was measured at 2 mm, indicating precise boundary delineation.
• The approach demonstrated efficient user control and reduced the need for post-segmentation refinement.

Conclusions:

• The interactive fuzzy connectedness method presents a practical and effective solution for lung delineation in dynamic 4D thoracic MRI for pediatric TIS.
• This technique offers a balance between user interaction and automated processing, yielding accurate and repeatable results.
• The findings support the utility of this method for quantitative image analysis in pediatric respiratory disease research.