(Tuesday, 7/31/2018) 11:00 AM - 12:15 PM
Room: Room 202
Purpose: For a health system with numerous SPECT systems, it is desired to have an efficient and effective method to review routine QC images (both 2D and 3D) from them and detect potential issues as early as possible.
Methods: An automated workflow of QC image review is built in our department. The workflow automatically processes QC images from PACS, identifies those with potential issues and notifies physics personnel by email. QC images sent to the workflow include daily flood field images (2D) and quarterly ACR SPECT phantom images (3D).NPS analysis for flood field images uses the previously proposed structured noise index (SNI) metric, whereas that for SPECT uniformity images uses a newly proposed loss of correlation (LOC) metric. Unlike flood field images, SPECT uniformity images are expected to have radial NPS profiles that peak at a certain spatial frequency. The correlation between the measured radial NPS profile from a SPECT uniformity image and the expected profile therefore carries information about how much a SPECT uniformity image deviates from its ideal performance. The loss of correlation (LOC), defined as the decrement from the ideal correlation of 1, is used as the metric in our automated workflow. QC images for five SPECT systems over a three-year period are retrospectively analyzed to demonstrate the validity of our method.
Results: The utility of the newly proposed LOC metric is demonstrated by quantitative results from test SPECT images without and with ring artifacts. Retrospective review of QC images with the automated workflow demonstrate good correlation with results from expert review.
Conclusion: Our experience in developing an automated workflow for reviewing QC images from SPECT system is presented. A newly proposed metric shows promises in the evaluation of SPECT uniformity.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
