Program Information
A Pilot Study Evaluating the Robustness of Density Mask Scoring (RA-950), a Quantitative Measure of Chronic Obstructive Pulmonary Disease, to CT Parameter Selection Using a High-Throughput, Automated, Computational Research Pipeline
J Hoffman*, H Kim , J Goldin , M Brown , M McNitt-Gray , UCLA School of Medicine, Los Angeles, CA
Presentations
SU-K-201-9 (Sunday, July 30, 2017) 4:00 PM - 6:00 PM Room: 201
Purpose: Previous robustness studies have been limited to one or two imaging parameters, such a slice-thickness, or slice-thickness and kernel. In this study, we conduct a multivariate, three-parameter, investigation into the robustness of RA-950 scoring to CT acquisition and reconstruction parameter selection in lung screening using novel, efficient, high-throughput, infrastructure.
Methods: Ten lung screening scans were identified from an in-house database. Reduced-dose acquisitions were simulated at 50% and 25% of the clinical dose, and reconstruction was performed using slice-thicknesses of 0.6, 1.0, 2.0, 5.0mm, and smooth, medium and sharp reconstructions kernels. 36 parameter configurations were reconstructed per patient; previous studies have been limited to, at most, 20 configurations. Reconstructions were performed using open-source weighted-filtered-backprojection (FreeCT_wFBP). RA-950 scoring was performed on each reconstruction and change in RA-950 relative to a reference configuration (100% dose, 1.0mm, medium kernel) was calculated. Changes were averaged across the population. Additionally, patient-specific surface plots of RA-950 as a function of parameter configuration were generated. Reconstruction and analysis were performed using an in-house, high-throughput, automated imaging and analysis “pipeline.”
Results: With the exception of the sharp reconstruction kernel, RA-950 scoring was observed to be fairly robust to dose and slice-thickness changes; most scores fell within +-5% of the score at reference. Protocols resulting in high image noise caused RA-950 scores to increase, agreeing well with the current literature. Surface plots revealed an effect on RA-950 scoring due to the amount of emphysema present in the patient. Reconstruction and analysis of 360 unique datasets took approximately one day, a substantial performance increase over previous, similar work.
Conclusion: RA-950 scoring for clinical use may need less strict protocol control than previously expected. Multivariate analysis techniques reveal the strong dependence of quantitative metrics on multiple parameters. High throughput infrastructure allowed for data generation and analysis in new, powerful ways.
Funding Support, Disclosures, and Conflict of Interest: Disclosures: *All authors: Institutional research agreement, Siemens Healthineers Funding support: *President Tobacco-Related Disease Research Program (UCOP-TRDRP grant #22RT-0131) *National Cancer Institutes Quantitative Imaging Network (QIN grant U01-CA181156).
Contact Email: