Figure 4. p-Values from Larsson et al Figure 4. p-Values from Larsson et al.<sup>(9)</sup> comparing 6 imaging conditions (3 imaging systems, each system processed using two different techniques) for thoracic imaging (5 test conditions vs a reference condition), using 6 observers, 6 image quality criteria and 20 cases per condition (non-paired data, except for image quality criterion/imaging condition combinations 1–6 where the compared images were paired). In the original study, observers were pooled and the p-values were calculated by two-sided z-tests of the ROCFIT output. In the reanalysis using VGC Analyzer, observers were treated either as fixed or random. The x-axis corresponds to the 30 image quality criterion/imaging condition combinations. Resulting p-values >0.20 have been excluded from the diagram. From: THE VALIDITY OF USING ROC SOFTWARE FOR ANALYSING VISUAL GRADING CHARACTERISTICS DATA: AN INVESTIGATION BASED ON THE NOVEL SOFTWARE VGC ANALYZER Radiat Prot Dosimetry. 2016;169(1-4):54-59. doi:10.1093/rpd/ncw035 Radiat Prot Dosimetry | © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Figure 3. AUC<sub>VGC</sub> and CI from Carlander et al Figure 3. AUC<sub>VGC</sub> and CI from Carlander et al.<sup>(8)</sup> comparing two dose levels for neonatal chest imaging (80 vs 100 %), using 5 observers, 4 criteria and 24 cases per condition (paired data). In the original study, observers were pooled and ROCFIT was used to determine the AUC<sub>VGC</sub> and the CI for each criterion. Hence, the paired-data effect could not be taken into account. In the reanalysis using VGC Analyzer, the paired-data effect was taken into account in the analysis and observers were treated either as fixed or random. From: THE VALIDITY OF USING ROC SOFTWARE FOR ANALYSING VISUAL GRADING CHARACTERISTICS DATA: AN INVESTIGATION BASED ON THE NOVEL SOFTWARE VGC ANALYZER Radiat Prot Dosimetry. 2016;169(1-4):54-59. doi:10.1093/rpd/ncw035 Radiat Prot Dosimetry | © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Figure 2. AUC<sub>VGC</sub> and CI from Carlander et al Figure 2. AUC<sub>VGC</sub> and CI from Carlander et al.<sup>(7)</sup> comparing two CR systems for neonatal chest imaging (double-side readout technique vs single-side readout technique), using 5 observers, 4 criteria and 24 cases per condition (non-paired data). In the original study, observers were pooled and ROCFIT was used to determine the AUC<sub>VGC</sub> and the CI for each criterion. In the reanalysis using VGC Analyzer, observers were treated either as fixed or random. All AUC<sub>VGC</sub> values were determined using binormal curve fitting. From: THE VALIDITY OF USING ROC SOFTWARE FOR ANALYSING VISUAL GRADING CHARACTERISTICS DATA: AN INVESTIGATION BASED ON THE NOVEL SOFTWARE VGC ANALYZER Radiat Prot Dosimetry. 2016;169(1-4):54-59. doi:10.1093/rpd/ncw035 Radiat Prot Dosimetry | © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Figure 1. (a) AUC<sub>VGC</sub> and CI from Zachrisson et al.<sup>(6)</sup> comparing two tube voltages for urography (55 vs 73 kV), using 5 observers, 3 criteria and 31 cases per condition (non-paired data). On the 55 kV images, the effective dose was reduced to 32 %, compared with the 73 kV images, by software simulation<sup>(11)</sup>. In the original study, observers and criteria were pooled and ROCFIT was used to determine the AUC<sub>VGC</sub> and the CI. In the reanalysis, using VGC Analyzer, criteria were pooled, but the observers were treated either as fixed or random. (b) AUC<sub>VGC</sub> and CI from Zachrisson et al.<sup>(6)</sup> comparing two tube voltages for urography [55 vs 73 kV (matched effective dose)], using 5 observers, 3 criteria and 31 cases per condition (non-paired data). In the original study, observers and criteria were pooled and ROCFIT was used to determine the AUC<sub>VGC</sub> and the CI. In the reanalysis using VGC Analyzer, criteria were pooled, but observers were treated either as fixed or random. From: THE VALIDITY OF USING ROC SOFTWARE FOR ANALYSING VISUAL GRADING CHARACTERISTICS DATA: AN INVESTIGATION BASED ON THE NOVEL SOFTWARE VGC ANALYZER Radiat Prot Dosimetry. 2016;169(1-4):54-59. doi:10.1093/rpd/ncw035 Radiat Prot Dosimetry | © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com