[Segmenting lung fields in serial chest radiographs using both population and patient-specific shape statistics].

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Accurate segmentation of lung fields in serial chest radiographs is crucial for monitoring respiratory conditions.
• Existing deformable models often struggle with patient-specific anatomical variations and limited initial data.

Purpose of the Study:

• To develop and evaluate a novel deformable model for robust lung field segmentation.
• To integrate both population-based and patient-specific shape statistics for enhanced accuracy.
• To improve adaptability to individual patient anatomy in serial radiographic analysis.

Main Methods:

• A modified scale-invariant feature transform (SIFT) local descriptor was employed to characterize image features.
• A deformable model was developed, guided by SIFT descriptors for region seeking.
• The model's deformation was constrained by a combination of population-based and patient-specific shape statistics, with adaptive weighting.

Main Results:

• The proposed model demonstrated adaptability to diverse patient lung field shape variability.
• The integration of both statistical approaches led to more robust and accurate segmentation outcomes.
• The model effectively balanced population-level priors with individual patient data over time.

Conclusions:

• The novel deformable model offers a significant advancement in lung field segmentation from serial chest radiographs.
• Combining population and patient-specific shape statistics enhances segmentation accuracy and robustness.
• This approach holds promise for improved clinical monitoring and analysis of respiratory diseases.