Creating a virtual ultra low-dose NLST data set and assessing the impact of dose reduction on nodule detection and characterization
We propose to apply our simulation technique to NLST data set to create a virtual ultra low dose dataset and assess the nodule detection and characterization performance according to varying dose reduction rate.
A subset of subjects will be randomly selected and downloaded from NLST database and will be used to create a virtual ultra low-dose NLST dataset.
Dose rates of 50%, 25%, 10% will be used. Three board certified radiologists will participate in reader study for lung nodule detection and characterization with ultra low-dose NLST dataset. Detection performance will be assessed with ROC analysis.
CT examinations in lung cancer screening need to be performed according to the ALARA principle. Yet, it is difficult to determine an appropriate dose level in which radiation dose in CT examinations is reduced maximally while diagnostic performance is not compromised. Our aim is to determine an appropriate dose level for nodule detection and characterization in asymptomatic screening population using NLST dataset and applying low-dose simulation technique.
-
Deep learning-enabled accurate normalization of reconstruction kernel effects on emphysema quantification in low-dose CT.
Jin H, Heo C, Kim JH
Phys Med Biol. 2019 Jul 4; Volume 64 (Issue 13): Pages 135010 PUBMED