Assessment of Interobserver and Intraobserver Variabilities in the Measurements of Anteroposterior (AP) and Lateral (LAT) Diameters on Paediatric CT images for Size-Specific Dose Estimation

Interobserver and Intraobserver Variabilities in Individual Size Measurements

Abstract

Introduction Both interobserver and intraobserver variability evaluations are important to validate the data reliability and ensure consistency in measurements. to assess the variability in anteroposterior (AP) and lateral (LAT) diameter measurements of the head and abdomen for determination of effective diameter among paediatric population. The measurements were conducted by various observers.
Methods Fifty CT images of paediatric head and abdomen were randomly selected from from five different age groups of paediatric patients, all aged below 18 years old. All images were acquired using the standard exposure protocol for head and abdomen CT scanning. The anteroposterior (AP) and lateral (LAT) diameters were measured on selected CT images of the head and abdomen at specific levels. To ensure measurement consistency, diameters were consistently measured at defined positions: the middle slice for the head and a position between the kidneys or the second lumbar vertebra for the abdomen. All diameter measurements were conducted by three observers and repeated three times. Statistical analysis, including the intraclass correlation coefficient (ICC) and Bland-Altman method for intraobserver and interobserver agreement, were employed to assess measurement consistency.
Results The findings showed excellent reliability in all measurements (LCC values > 0.9) for both interobserver and intraobserver. The variability between student, physicist and radiologist was high. The deviation of observer measurements between all observers were also significant for AP and LAT diameters of head and abdomen of the selected CT images.
Discussions The findings resulted in high reliability, showing strong agreement among measurements and low variability in measurements among all observers and amongst all participating centres. The variability between the principal investigators was lower, but still substantial. The deviation of observer measurements from the gold rater were also significant for all CT parameters.

Conclusion We found excellent reliability for both intra- and interobserver measurements.