Respiratory Invariant Textures From Static Computed Tomography Scans for Explainable Lung Function Characterization.

Area of Science:

• Radiology and Medical Imaging
• Pulmonary Medicine
• Computational Pathology

Background:

• Lung tissue characteristics independent of breathing can inform function assessment.
• Computed tomography (CT) and ventilation scans offer complementary data for lung evaluation.
• Radiomics aims to extract quantitative features from medical images for clinical insights.

Purpose of the Study:

• To correlate textural signatures from 4-dimensional CT scans with pulmonary function measurements.
• To identify specific radiomic features associated with lung function in cancer patients.
• To explore the spatial agreement and robustness of these features.

Main Methods:

• Utilized 4D CT, DTPA-SPECT ventilation scans, and spirometry from 21 lung cancer patients.
• Identified 8 function-correlated radiomic features from 79 candidates based on statistical significance.
• Generated feature maps (FMs) and validated using Spearman correlation, Dice similarity, and intraclass correlation coefficients.

Main Results:

• Eight function-correlated radiomic features were identified at the subregion level.
• Feature maps showed moderate-to-strong correlations with ventilation scans.
• The gray level dependence matrix dependence nonuniformity feature demonstrated robust spatial correlation and predicted FEV1 and FEV1/FVC.

Conclusions:

• Pulmonary textures identified via CT are associated with local and global lung function.
• These findings offer insights into the link between lung texture and function.
• Further validation is needed for clinical application of these radiomic features.