Test-Retest Reliability Assessment of Computed Tomography Subjective and Objective Resolution under Differentiated Scanning Parameters

Purpose Based on existing CT performance testing methods, the test-retest reliability of subjective and objective results for both spatial resolution and low-contrast detectability was evaluated and analyzed under different scanning conditions. Methods A prospective research method was adopted, using clinical routine scanning protocols and parameters for the head, chest and abdomen. Eight CT scanners from different regions underwent two rounds of test, with each round tested three times. The coefficient of variation (CV) and intraclass correlation coefficient (ICC) were used to evaluate inter-CT test-retest reliability. Bland-Altman analysis was used to evaluate intra-CT test-retest reliability. Results For head scans, the CV range for spatial resolution was 8.85%~21.66%, with ICCs from 0.26 to 0.72; the CV range for low-contrast detectability was 5.53%~47.84%, with ICCs from 0.16 to 0.75. For chest scans, the CV range for spatial resolution was 9.13%~19.65%, with ICCs from 0.41 to 0.75; the CV range for low-contrast detectability was 12.25%~38.89%, with ICCs from 0.18 to 0.68. For abdomen scans, the CV range for low-contrast detectability was 5.91%~37.33%, with ICCs from 0.23 to 0.77. For the same CT, the successive results of subjective spatial resolution from head protocols demonstrated statistical significance under the first enhancement strength. Moreover, the successive results of objective spatial resolution from chest protocols demonstrated statistical significance under the first, second and third enhancement level. No statistical significance was observed under other scanning conditions. Conclusion The objective contrast evaluation method demonstrated better variability and consistency compared to the subjective evaluation method under most test conditions. Even for the same CT with iterative reconstruction algorithm at the same level, there may exist statistical significance of the spatial resolution values from successive head and chest scanning. Cautions should be taken when test results of objective spatial resolution are analyzed.