Room: AAPM ePoster Library
Purpose: This study compares the contrast between two Canon iterative reconstruction algorithms: AIDR 3D and the new model-based iterative reconstruction algorithm, FIRST. Two metrics are used: the traditional contrast-to-noise ratio (CNR), and the low-contrast object specific CNR (CNR(LO)). CNR(LO) evaluates the impact of noise texture on size-specific object detectability. Dose levels and reconstruction settings appropriate for lung screening, abdomen, and brain studies were investigated.
Methods: A Canon Aquilion Genesis scanner was used to scan the ACR image quality phantom at dose levels of 2.8 mGy, 19.7 mGy, and 29.4 mGy (reported in the 32-cm phantom), which are clinically-appropriate for lung screening, abdomen, and brain exams, respectively. Each scan was repeated 10 times and reconstructed using AIDR 3D and FIRST with kernel settings appropriate for lung, body, and brain exams. CNR and CNR(LO) were calculated for the 25-mm and 6-mm low-contrast disks from 10 1-mm slices on each of the 10 scans, for a total of 100 measurements per dose/algorithm/kernel combination. CNR(LO) was calculated by replacing the standard deviation in the CNR equation by the square root of the noise power spectrum evaluated at the disk’s spatial frequency. A two-tailed t-test was used to assess significant differences between FIRST and AIDR 3D using each metric.
Results: Using the traditional CNR metric, FIRST outperformed AIDR 3D for all scans. Using the CNR(LO) metric, however, FIRST performed significantly better than AIDR 3D only for abdomen scans with the 6-mm contrast disk. Results from the abdomen scan with 25-mm disk were not significantly different, and FIRST produced a significantly lower CNR(LO) than AIDR 3D for lung and brain scans with both contrast disks.
Conclusion: FIRST demonstrates improved contrast over AIDR 3D for phantom-simulated abdomen scans, especially for smaller low-contrast test objects. AIDR 3D performs better for simulated lung screening scans and brain scans.
Funding Support, Disclosures, and Conflict of Interest: This work was funded by Canon Medical Systems USA, Inc.