Program Information
Noise-Robust and Accurate Measurement Method of Small Airway Wall Thickness in Low-Dose CT Scan: A COPDgene Phantom Study
Z Yang*, H Jin , J Kim , Seoul National University, Seoul, KOREA, Republic of
Presentations
SU-F-207-4 (Sunday, July 12, 2015) 4:00 PM - 6:00 PM Room: 207
Purpose:
We present a novel method, attenuation profile matching (APM) technique, which provides accurate determination the wall thickness of small airways as well as superior robustness to noise and reconstruction kernel variation.
Methods:
Point spread functions (PSF) of a commercial CT (Sensation 16, Siemens) were acquired by using a wire phantom for various reconstruction kernels (B10s ~ B80s). Sets of COPDgene phantom CT data were created for varying exposure (25 to 200 mAs) and FOVs, and with different reconstruction kernels. Line profiles were obtained across the airway walls with varying sizes (0.6 - 1.5mm), and compared with synthetic line profiles which were generated by taking convolution of the PSF with a the numerical airway wall phantoms derived from the variation of COPDgene phantom design data. Wall thickness of a given airway was determined as that of the numerical model yielding minimum error. The performance of APM method was compared with that of two other methods (FWHM, IBM) for varying exposure conditions and with different reconstruction kernels.
Results:
Measurement error the APM method ranged from 0.01 to 0.15 (± 0.012) mm, whereas that of the FWHM and IBM method was from 0.07 to 0.88 (± 0.015) mm and 0.08 to 0.72 (± 0.015) mm, respectively, for varying range of wall thickness and different reconstruction kernels. Average error of APM method was markedly lower (p<0.01) than that of FWHM and IBM methods.
Conclusion:
Our proposed APM technique could determine the airway wall thickness accurately even for the small airways less than 1mm. Our technique may allow detection of early change of airway wall thickening in early diagnosis, patient sub-typing, and therapeutic monitoring in the management of COPD disease.
Contact Email: