Abstract
Accurate characterization of lung tissue in thoracic computed tomography (CT) is essential for the early diagnosis, phenotyping, and personalized treatment of pulmonary diseases. However, quantitative assessment of lung parenchymal abnormalities is challenging because parenchymal density variations are confounded by inter-subject anatomical differences. To address this problem, we propose a parametric statistical model based on principal component analysis (PCA) of lung parenchymal density derived from volumetric CT scans. The method registers a CT cohort to a reference subject and disentangles anatomical and parenchymal variability through a statistical framework controlled by a single hyperparameter. This parameter enables the identification of subject-specific parenchymal density variations while minimizing the influence of anatomical differences. By training regression functions, we demonstrate that the resulting model captures clinically relevant patterns of parenchymal abnormality and exhibits significant correlations with quantitative imaging biomarkers and conventional pulmonary function measurements. In particular, the model is sensitive to ground-glass opacities associated with interstitial lung disease. For reproducibility and further research, the learned PCA model, the variables, and the learned regressors are made publicly available for academic use.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/TIA_001.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: Not Available
BibTex
@InProceedings{EstVic_Quantification_MICCAISAT2026,
author = { Estopier Castillo, Vicente AND Franco, Jean-Sébastien AND Cohen, Julien AND Bayat, Sam AND Pujades, Sergi},
title = { { Quantification of Parenchymal Abnormalities Using a Statistical Model of Lung CT Density } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17269},
month = {pending},
page = {pending}
}