Reproducibility of linear tumor measurements using PACS: comparison of caliper method with edge-tracing method

The aim of this study was to evaluate inter- and intra-observer reproducibility when making electronic caliper linear tumor measurements on picture archiving and communications systems (PACS) and compare them with linear measurements obtained from circumferential tracing of tumor perimeter. Three radiologists measured 64 masses from 30 patients on body CT scans in two separate settings. Long axis and perpendicular short axis were measured using electronic calipers. The edge of each tumor was traced electronically and the long and short axes were calculated by computer software. The reproducibility of a measurement was evaluated by computing and comparing the absolute value of the mean difference between initial and subsequent measurements. The mean differences +/-95% confidence interval (CI) between two measurements of the long by short axis were 3.8+/-2.6x3.1+/-1.8 mm when the caliper method was used and 3.5+/-2.0x3.2+/-1.5 mm when the tumor tracing method was used. There was no statistically significant difference in individual intra-observer reproducibility of tumor axes measurements. Neither long- nor short-axis single-dimension measurements resulted in significantly greater or lesser intra-observer reproducibility. When comparing caliper and tracing measurements, the overall mean difference (3.42+/-1.8 vs 3.38+/-1.4 mm) was not statistically significant. There was close correlation between the individual measurements made by each observer whether these were made by electronic calipers and when these were calculated from electronic tracings (Pearson correlations between 0.79 and 0.949). Current PACS systems allow reproducible linear, long or short axis, tumor measurements. There is no significant difference in reproducibility of measurements whether these are made directly with electronic calipers or calculated from tumor edge tracings.