Encrypted login | home

Program Information

Evaluation of High Contrast Resolution for Model Based Iterative Reconstruction of Sinus Examinations

R Paden

R Paden*, W Pavlicek, Mayo Clinic Arizona, SCOTTSDALE, AZ

SU-E-I-54 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall

Purpose: To evaluate high contrast resolution of Model Based Iterative Reconstruction used with typical sinus examination acquisition parameters.

Methods: MBIR has recently become available as a recon option on a clinical scanner (Discovery HD750, GE Healthcare). In this work, we evaluate high contrast resolution for scan and reconstruction options that are available for sinus examinations. For this study we used our adult sinus CT protocol reconstructed with filtered back projection, and two alternative scans reconstructed with IR. Our conventional adult sinus CT protocol utilizes a high resolution scan mode which is not compatible with the scanner's IR recon option. The two additional scans are with high resolution option off, one with the head SFOV, and one with the body SFOV. Using IR and the head SFOV, reconstructed images have a 1024 x 1024 pixel matrix. Using IR and the body SFOV, reconstructed images have a 512 x 512 pixel matrix. Three evaluations of high contrast resolution are made for these images. A wire phantom is scanned for assessment of image modulation transfer function. The bar patterns of the ACR phantom are visually assessed for quality in both axial and coronal reformats.

Results: MTF curves show 50% values of 6.8, 7.5, and 7.7 lp/cm for Body IR 512 x 512, Head IR 1024 x 1024, and filtered back projection with HD Bone kernel. The 10% MFT curves for these reconstructions are 11.2, 11.9, and 12.1 lp/cm. Visual evaluation of the ACR phantom at 15 cm display field of view demonstrates resolution of the 10 lp/cm bar patterns for all reconstructions with better visualization of the axial versus coronal recons.

Conclusions: MBIR reconstruction demonstrates high contrast resolution that is comparable with our conventional sinus examination.

Contact Email