# 4D-CT strain analysis for biomechanical characterization of COPD in patients undergoing radiotherapy planning: an exploratory study

## TL;DR

This study explores using 4D-CT scans to assess lung function in COPD patients, showing that a strain parameter called Speedmax can help distinguish COPD from normal lung function.

## Contribution

The study introduces 4D-CT strain analysis as a novel method for biomechanical characterization of COPD during radiotherapy planning.

## Key findings

- Speedmax was significantly reduced in COPD patients and strongly correlated with pulmonary function indices.
- Speedmax demonstrated good diagnostic performance with high sensitivity and moderate specificity for detecting COPD.
- Multivariate analysis confirmed Speedmax as an independent predictor of COPD status after adjusting for tumor size.

## Abstract

Chronic obstructive pulmonary disease (COPD) is characterized by progressive airflow limitation and altered pulmonary biomechanics. While conventional spirometry provides global functional information, it cannot assess regional heterogeneity. This exploratory study aimed to evaluate the diagnostic performance of four-dimensional computed tomography (4D-CT) strain analysis in differentiating normal lung function from COPD in a cohort of patients with small pulmonary lesions who underwent 4D-CT as part of radiotherapy planning, thereby establishing feasibility for potential future integration into treatment workflows.

This single-center retrospective cross-sectional study included 46 patients with pulmonary lesions ≤ 3 cm who underwent 4D-CT localization scanning for radiotherapy planning between January 2021 and March 2024. Patients were stratified into normal lung function (n = 34; FEV1/FVC ≥ 0.7 and FEV1 ≥ 80% predicted) and COPD groups (n = 12; FEV1/FVC < 0.7) based on spirometric results according to GOLD criteria. Strain parameters—maximum principal strain (PSmax), mean principal strain (PSmean), and maximum displacement velocity (Speedmax)—were calculated using optical flow algorithms. Internal validation using bootstrap resampling (1,000 iterations) and multivariate analysis adjusting for tumor size were performed. All lung parenchyma, including tumor regions, was included in the strain analysis to reflect real-world clinical scenarios.

Despite the small COPD sample size (n = 12), the study achieved adequate statistical power (88.1%) due to large effect sizes. COPD patients exhibited significantly higher symptom prevalence and severely impaired pulmonary function (FEV1/FVC: 58.44 ± 8.78% vs. 82.31 ± 6.41%, P < 0.001, Cohen’s d = 3.373). Speedmax demonstrated statistically significant reduction in COPD patients (763.22 ± 85.22 vs. 1074.19 ± 319.68 mm/s, P < 0.001, d = 1.114), while PSmax and PSmean showed trends toward reduction. Speedmax demonstrated strong correlations with pulmonary function indices, particularly with FEV1/FVC (r = 0.533, P < 0.01), FEV1 (r = 0.445, P < 0.01), and MEF50 (r = 0.525, P < 0.01). Receiver operating characteristic analysis revealed good diagnostic performance for Speedmax (AUC = 0.886, optimism-corrected AUC = 0.889, sensitivity = 100.0%, specificity = 72.7%). Multivariate analysis confirmed that Speedmax remained a significant independent predictor of COPD status after adjusting for tumor size (AUC = 0.931). Bootstrap internal validation showed minimal optimism bias (-0.001), confirming model robustness.

This exploratory study provides preliminary evidence that 4D-CT strain analysis, particularly Speedmax, shows promise for differentiating normal lung function from COPD in patients with small pulmonary lesions. However, given the small sample size and single-center design, diagnostic performance metrics should be interpreted with caution. External validation in larger, multi-center cohorts is required before clinical implementation.