Correction of lung boundary using the gradient and intensity distribution

Area of Science:

• Medical Imaging
• Image Segmentation
• Computational Anatomy

Background:

• Accurate segmentation of lung boundaries in computed tomography (CT) is essential for quantitative analysis, particularly in distinguishing between expiratory and inspiratory phases.
• Existing segmentation methods can struggle with accuracy, leading to potential errors in volume measurements and disease assessment.

Purpose of the Study:

• To develop and validate a new computational method for correcting inaccurate lung boundaries in CT images.
• To enhance the reliability and reproducibility of lung segmentation in both expiratory and inspiratory CT scans.

Main Methods:

• Initial lung boundary extraction using density-based segmentation.
• Computation of boundary propagation scope via gradient profile analysis with adaptive length to prevent leakage and improve efficiency.
• Boundary propagation using a gradient and intensity-based speed function to avoid convergence to local maxima.

Main Results:

• The proposed method demonstrated reliable and reproducible correction of segmented lung boundaries.
• Evaluation through visual inspection, accuracy metrics, and processing time confirmed the effectiveness of the approach.
• The adaptive scope definition successfully prevented segmentation leakage and enhanced propagation efficiency.

Conclusions:

• The novel lung boundary correction method offers a robust solution for improving CT image segmentation accuracy.
• This technique has the potential to enhance the reliability of quantitative lung analysis in clinical and research settings.