| Popis: |
Purpose: The purpose of this thesis is to understand the relationship between mammographic performance based on actual clinical reading and performances at screen read test-sets, as well as to examine the potential causal agents for any lack of correlation. Methods: This study was designed to encompass three facets. The first element investigated the extent to which test-set reading can represent actual clinical reporting in screening mammography. The second element examined the manner that the location where reading takes place and the availability of prior images can impact upon performance in breast test-set reading. The third element considered the reading workstation monitors and the viewing environment available within BreastScreen New South Wales centres to determine whether consistent reporting conditions were provided to breast screen readers. Results: Moderate or acceptable level of agreement (W = 0.69–0.73, P < 0.01) were shown between actual clinical reporting and test-set conditions when describing group performance. The agreement was enhanced when prior images were available. The location where reading takes place and the availability of prior images showed acceptable levels of agreement (W = 0.75–0.79, P < 0.001) between group performance although both factors had a varying impact when examining the results of individual reader. The final aspect demonstrated an overall good adherence of reading workstation monitors and the viewing environment to published guidelines. Conclusions: Test-set readings in clinical and laboratory settings can be used to represent radiologic group performance in the clinic to a reasonable level particularly if prior images are available. If individual efficacy is being examined, some observers do demonstrate differences between test-sets and clinical performance, as well as differences between test-set situations even when viewing conditions are generally adhering to international standards. |
