Precise Lung Density Quantification with a Physics-based CT Harmonizer

Area of Science:

• Medical Imaging
• Radiology
• Pulmonary Medicine

Background:

• Chest CT scans exhibit variability in noise, spatial resolution, and lung volume.
• These variations impact the reproducibility of lung density measurements, crucial for diagnosing and monitoring diseases like Chronic Obstructive Pulmonary Disease (COPD).

Purpose of the Study:

• To develop and evaluate a physics-based image harmonization method for CT scans.
• The goal is to standardize images to a reference quality index, improving lung density measurement reproducibility.

Main Methods:

• A retrospective analysis of chest CT data from the COPDGene study was performed.
• A harmonization algorithm adjusted spatial resolution, noise, and lung volume sequentially.
• Lung density metrics, including LAA-950 and Perc15, were calculated and compared to existing methods (VALD, MF-VALD).

Main Results:

• The harmonization technique significantly improved Perc15 reproducibility by 4.8-fold.
• It outperformed existing methods, achieving better reproducibility coefficients on both full-dose and reduced-dose scans.
• The harmonizer reduced Perc15 variability from 35.6 HU to 7.4 HU.

Conclusions:

• The developed physics-based technique effectively harmonizes CT images to a reference quality index.
• This harmonization improves the reproducibility of lung density metrics, aiding in more reliable disease assessment.
• The method offers a significant advancement for quantitative CT analysis in pulmonary research and clinical practice.