Toward Robust Characterization of Lung Diseases: A Sensitivity Analysis of Lung Computed Tomography Biomarkers to Registration Error

Abstract

Computed tomography (CT) scans, because of their ability to differentiate tissue densities, have been widely used to evaluate lung health. Recent studies such as COPDGene have collected inhalation and exhalation CT scans from thousands of subjects, promising insight into the mechanical properties of lung tissue. These paired scans must often be spatially aligned (i.e., registered) to extract biomarkers describing the movement of lung tissue that may correlate with disease. Unfortunately, the relationship between registration and biomarker error is poorly characterized, a challenge that must be addressed before registration-based biomarkers can be used in clinical practice. In our analysis, we consider three registration-based biomarkers (Jacobian determinant, anisotropic deformation index, and slab-rod index) and demonstrate their sensitivity to modeled registration error. We provide a range of errors for a given biomarker, highlighting how both the magnitude of registration error and correlations between vectors in the registration error field can influence biomarker error. We then describe a method to measure the error field for a particular registration algorithm and compare it with modeled registration error. These estimates enable selection of an appropriate registration error model, which improves understanding of biomarker uncertainty. Quantifying the relationship between registration and biomarker error is crucial because it may inform the selection of a registration algorithm to reduce error in new research studies, and in turn, result in robust imaging biomarkers for disease characterization.