WCLC 2025 - Posters & ePosters-P1.04.13 The Impact of Reconstruction Kernel and Hounsfield Unit Threshold on Emphysema Quantification From Lung Cancer Screening Data

P1.04.13 The Impact of Reconstruction Kernel and Hounsfield Unit Threshold on Emphysema Quantification From Lung Cancer Screening Data

Back to course

Pdf Summary

This study investigates how different CT reconstruction kernels and Hounsfield Unit (HU) thresholds affect emphysema quantification using low attenuation area (LAA) analysis in lung cancer screening CT scans. Emphysema, a progressive and irreversible lung disease, can be assessed early via quantitative computed tomography (QCT) by analyzing areas of low lung density. However, CT parameters like the reconstruction kernel and HU threshold can influence measurement outcomes.<br /><br />Using a subset from the Netherlands-China Big-3 (NELCIN B3) dataset of 352 scans, with 266 cases agreed upon by four radiologists, the study applied AI-based software (Aview v.42) to measure the percentage of lung area classified as low attenuation (LAA%) under varying kernel and HU threshold settings. Statistical tests including Friedman and Wilcoxon sign tests (with Holm correction) were used to analyze differences.<br /><br />Results showed that the chosen CT kernel had a much larger impact on LAA% than the HU threshold. Specifically, the median LAA% using kernel D45f was about ten times higher than that from kernels B30f and B80f across all HU thresholds. Differences between B30f and B80f kernels were not statistically significant at the same HU level. Additionally, LAA% distributions showed substantial overlap between cases with and without radiologist-diagnosed emphysema, indicating that LAA analysis alone does not clearly differentiate emphysema presence.<br /><br />The study concludes that both reconstruction kernel and HU threshold significantly affect emphysema quantification from CT, but current quantitative methods lack sensitivity and specificity to match radiologic visual assessments. Thus, further refinement of quantification algorithms is required to improve diagnostic accuracy. This highlights the importance of standardizing CT imaging parameters and enhancing AI analysis tools for effective emphysema screening using lung cancer CT data.

Asset Subtitle

Hailan Liu

Meta Tag

Speaker Hailan Liu

Topic Screening and Early Detection

Keywords

CT reconstruction kernels

Hounsfield Unit thresholds

emphysema quantification

low attenuation area analysis

lung cancer screening

quantitative computed tomography

NELCIN B3 dataset

AI-based software Aview v.42

statistical analysis Friedman Wilcoxon

diagnostic accuracy in emphysema