Hertel-exophthalmometry-like multi-detector-row-CT-exophthalmometry: inter-disciplinary inter-observer reproducibility of measurements

Abstract

Objectives: To investigate the interdisciplinary interobserver reproducibility of Hertel-exophthalmometry-like protrusion measurements on multidetector-row-computed-tomography- (MDCT-) images of the orbit to facilitate structured evaluation of the orbit and mid-face.

Methods: Respective reproducibility of base-length along the interfronto-zygomatic line, right and left ocular protrusion, and deriving interocular difference was measured in this retrospective (04/2009-03/2020) single-centre observational study. MDCT-series and slice-positions were selected independently, using picture-archiving-and-communication-system- (PACS-) tools on tilt-corrected axial MDCT-images (slice-thickness 0.6-3.0 mm, window/centre 350/50 HU) in 37 selected adult patients (24 female, age 57 ± 13 years, average±standard-deviation) with clinical indication for Hertel-exophthalmometry, by one radiology-attending, two ophthalmology-attendings, one critical-care-attending, and one ear-nose-throat-surgery resident, respectively. Bland-Altman plots and Wilcoxon-matched-pairs-signed-rank-tests compared interobserver results.

Results: Mean and median interobserver and intraobserver (radiology-attending) deviations were within 1 mm of respective averages of base-length (98 ± 4 mm), right and left ocular protrusion (21 ± 4 mm) and interocular difference (2 ± 1 mm). Relative interobserver deviations were within 2.0% of average (all patients) for base-length, and 5.0% (>80% of patients) for ocular protrusion. Pairwise interobserver comparison showed no significant differences between interocular differences of protrusion.

Conclusions: Respective measurements of base-length, ocular protrusion, and deriving interocular difference show high interdisciplinary interobserver reproducibility in tilt-corrected axial MDCT-images of the orbit or mid-face.

Advances in knowledge: Hertel-exophthalmometry-like protrusion measurements did not depend on the years of experience or the medical subspecialty of the observer. Measurements are objective, well reproducible and important for multiple medical disciplines and should thus be included in pertinent radiology reports.

Figure 1.

Axial multiplanar reconstruction, of unenhanced 64-row multidetector-row-computed-tomography of the orbits (slice thickness, 3 mm). Tilt-correction was deemed unnecessary. Two-sided arrow along the interfronto-zygomatic line demonstrates measurement of base-length. Perpendicular two-sided arrows show respective measurements of right and left ocular protrusion.

Figure 2.

Respective Bland-Altman plots show interdisciplinary interobserver agreement with upper and lower limits of agreement (dark grey horizontal lines) among five independent observers (O1, diamonds; O2, squares; O3, triangles; O4, cross-marks; O5, circles). Measurements include base-length (a, upper left panel), interocular difference of protrusion (b, upper right panel), right ocular protrusion (c, lower left panel) and left ocular protrusion (d, lower right panel) in 37 patients. Unit of measurement is 1 mm in each panel.

Figure 3.

Tilt-corrected multiplanar reconstructions of contrast-enhanced 64-row multidetector-row-computed-tomography of the orbits are orientated along the interfronto-zygomatic reference line as displayed by PACS (arrows in a, coronal reconstruction, and b, axial reconstruction; slice thickness, 2 mm).

Figure 4.

Tilt-corrected multiplanar reconstructions of contrast-enhanced 64-row multidetector-row-computed-tomography of the orbits are orientated along the interfronto-zygomatic reference line (slice thickness, 2.5 mm). A contrast-enhancing orbital mass displaces the left eyeball (arrow in a, coronal reconstruction), impeding exact measurement of left ocular protrusion on the axial reconstruction (b). This patient was excluded from the study because of incomplete data sets, i.e., Hertel-exophthalmometry had been attempted but failed.