SU-E-I-31: Image-Based Kernel Conversion Technique Normalizes the Reconstruction Kernel Effects in the Measurement of Emphysema Index in CT

Authors


Abstract

Purpose:

The emphysema index (EI) in CT is a quantitative measure of emphysema, which is known to be affected by reconstruction kernel. This study presents an image-based kernel conversion technique which converts CT image of sharp kernel to that of standard kernel and evaluates its impact on EI normalization for images obtained with different kernels.

Methods:

Sixty cases of CT exams obtained with 120kVp, 40mAs, 1mm thickness, of 2 reconstruction kernels (B30f, B50f) were selected from the low dose lung cancer screening database of our institution. An image-based kernel conversion technique, which converts by applying a conversion function, was performed to the CT data set of B50f to create a converted B30f data set. The conversion function(Fcon) which is the ratio of the spectral components of target and source kernel images, was produced using a training dataset. To test variability due to training dataset, two versions of conversion functions were produced using two different training dataset. The EI (RA950) was measured with a software package (Pulmonalizer, Seoul, South Korea) and compared for data sets of B50f, B30f, and the converted B30f. The accuracy of kernel conversion was evaluated with the mean and standard deviation of pair-wise differences in EI.

Results:

Population mean of EI was 28.89±6.48% for B50f data set, 10.96±6.37% for the B30f data set, and 11.27±6.83% for the converted B30f data set. The mean of pair-wise differences in EI between B30f and the converted B30f is 0.82% for Fcon1 and 1.07% for Fcon2. The correlation between the EI of two data sets was 0.99.

Conclusion:

Our study demonstrates the feasibility of image-based kernel conversion technique for normalization of kernel effect in measurement of EI. This technique has a potential to be used in evaluating the longitudinal changes of EI even when the CT was reconstructed with different kernels.

The research was supported by the Ministry of Trade, Industry & Energy (10043118)

Ancillary