Early COPD Risk Decision for Adults Aged From 40 to 79 Years Based on Lung Radiomics Features

Yingjian Yang, Wei Li, Yingwei Guo, Yang Liu, Qiang Li, Kai Yang, Shicong Wang, Nanrong Zeng, Wenxin Duan, Ziran Chen, Huai Chen, Xian Li, Wei Zhao, Rongchang Chen, Yan KangFrontiers in Medicine 2022 · DOI

Abstract

Once the age is above 60 years, the patients with COPD need to take action to prevent the progress and deterioration of COPD. Once the age is above 65 years, the patients with COPD stage 0 need to take precautions against COPD.

Introduction

Chronic Obstructive Pulmonary Disease (COPD) remains a leading cause of global mortality, defined by persistent and often progressive airflow limitation. The current diagnostic gold standard, Pulmonary Function Tests (PFTs), frequently fails to identify the disease until substantial, irreversible functional loss has occurred. A significant clinical challenge is the "aging lung"; as structural changes naturally occur with age, distinguishing between normal senescence and early-stage COPD becomes difficult. This study proposes a machine learning framework using lung radiomics to provide an early warning system, identifying critical age-stratified thresholds for clinical intervention.

Methodology

The researchers developed a sophisticated multi-stage pipeline to evaluate COPD risk. The process began with automated lung parenchyma segmentation using a ResU-Net deep learning architecture on High-Resolution Computed Tomography (HRCT) scans. From these segments, 1,316 quantitative radiomic features—including shape, intensity, and texture—were extracted using the PyRadiomics library.

To manage this high-dimensional data, the Lasso (Least Absolute Shrinkage and Selection Operator) method was applied to select the most predictive features. These features were then integrated into four distinct Survival Cox models designed to predict transitions between COPD stages across eight 5-year age intervals (40–79 years).

Workflow of the COPD risk evaluation framework across four distinct survival models (A-D).

The models were categorized as follows:

  • Model A: Transition from Stage 0 to any COPD stage.
  • Model B: Transition from Stage 0 to Stage I.
  • Model C: Transition from Stage 0 to Stage II.
  • Model D: Transition from Stage 0 to Stage III & IV.

Study Population and Datasets

The study analyzed a cohort of 468 Chinese subjects recruited from the National Clinical Research Center of Respiratory Diseases in Guangzhou. Subjects were categorized according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages, ranging from Stage 0 (at risk but no airflow limitation) to Stage IV (very severe).

Subject selection flow diagram illustrating inclusion/exclusion criteria and cohort distribution.

The dataset included 130 subjects at Stage 0, 109 at Stage I, 122 at Stage II, 74 at Stage III, and 33 at Stage IV. All participants underwent HRCT imaging to provide the raw data for radiomic analysis.

Key Insights and Results

The radiomics-based models demonstrated exceptional predictive performance. Model A achieved an Area Under the Curve (AUC) of 0.87 and a C-index of 0.94. Models predicting transitions to more severe stages (C and D) showed even higher accuracy, with AUCs of 0.94 and 0.97, respectively.

A primary finding of this research is the non-linear progression of COPD risk, which appears to "level up" in discrete 5-year increments. Radiomics serves as a "digital biopsy," uncovering mathematical textures in lung tissue that are invisible to the human eye but highly indicative of disease progression. For instance, original_shape_SurfaceVolumeRatio emerged as a critical biomarker for general COPD suffering, while texture features like ZoneEntropy were more predictive of advanced disease stages.

Clinical Implications

The study identifies clear age milestones for intervention:

  • Age 60: Patients already diagnosed with COPD must take aggressive action to prevent further deterioration.
  • Age 65: Individuals at Stage 0 (at risk) enter a "danger zone" where the probability of progressing to Stage I increases significantly every five years.
  • Age 70: The risk of progressing from Stage 0 to severe COPD (Stages II-IV) jumps sharply.

Conclusion

By leveraging high-dimensional imaging data and survival modeling, this framework provides a data-driven roadmap for early COPD management. While future work should address data imbalances in younger cohorts and explore non-linear deep learning survival models, this radiomics approach offers a precise tool for transitioning from reactive treatment to proactive, age-stratified prevention.

Summary updated

Abstract

Once the age is above 60 years, the patients with COPD need to take action to prevent the progress and deterioration of COPD. Once the age is above 65 years, the patients with COPD stage 0 need to take precautions against COPD.
BibTeX
@inproceedings{yang2022early,
  author = {Yingjian Yang and Wei Li and Yingwei Guo and Yang Liu and Qiang Li and Kai Yang and Shicong Wang and Nanrong Zeng and Wenxin Duan and Ziran Chen and Huai Chen and Xian Li and Wei Zhao and Rongchang Chen and Yan Kang},
  title = {{Early COPD Risk Decision for Adults Aged From 40 to 79 Years Based on Lung Radiomics Features}},
  booktitle = {Frontiers in Medicine},
  year = {2022},
  doi = {10.3389/fmed.2022.845286},
}
Figure