Algorithm variability in the estimation of lung nodule volume from phantom CT scans: results of the QIBA 3A public challenge
Adele P. Peskin, Alden A. Dima, Ganesh Saiprasad, Andrew Buckler
Quantifying changes in lung tumor volume is important for diagnosis, therapy planning, and evaluation of the response to therapy. Lung tumor volume change is determined by post-processing Computer Tomography (CT) scans of the lung, and good quantitative measurement is dependent upon consistency in both scanning procedures and post-processing procedures. The Quantitative Imaging Biomarker Alliance (QIBA) has defined standard procedures for measuring lung tumor volume changes in a document called a Profile, which defines standard working procedures for accurate and reproducible measurement of imaging biomarkers. The Profile is intended to reduce the variation of CT images across scanners and scanning environments. The aim of this study is to measure the variation of tumor volume calculations, to make sure that the overall process, both scanning and post-processing, can be accomplished according to the standards set by the QIBA Profile. Those standards specify that variation in nodule volumes should be less than 15% for solid nodules larger than 10 mm, reconstruction slice thickness ≤ 2.5mm, and densities > -630HU. Nodules outside of this range were also included in separate analyses to test the variability of volume measurements across a wider range of parameters.