Intra- and inter-observer measurement reliability using the Extend System measurement probes

David Schlesinger1, Zhiyuan Xu2, Chun-Po Yen1, Jason P. Sheehan, PhD3

1Charlottesville, United States 2University of Virginia 3Department of Neurosurgery, University of Virginia

Keywords: Extend, gamma knife, Fractionated radiosurgery, physics, technique

Interactive Manuscript

Ask Questions of this Manuscript:



     The Gamma Knife Extend System makes use of high-precision digital measurement probes to determine the patient’s position within the Extend frame. 
     This study investigates the degree of intraobserver and interobserver measurement repeatability and agreement on a phantom and a normal human subject.
     For the phantom test, the Extend Repositioning Check Tool (RCT) QA stand was used as a phantom. Three observers took measurements using the Extend digital measurement probes on every hole on the RCT three times in random order. The phantom procedure was repeated on a healthy human subject. A patient-specific frontpiece was created for the subject, and the subject was placed into setup position within the Extend frame. Two observers made five repeated measurements on three holes per RCT plate in random order. For both tests, repeatability coefficients (RC) were computed to describe intraobserver repeatability. Intraclass correlation coefficients (ICC)(2,1) were computed to describe the level of interobserver measurement interchangeability. Bland-Altman 95% limits of agreement (LOA) between observers were computed to quantify the expected measurement agreement.
     For the phantom measurements, the mean RC (mm) per plate were anterior(A)=0.022, superior(S)= 0.006, left/right(L/R)=0.099. The ICC were: (A)=1.000, (S)=0.981, (L/R): 0.544. The mean LOA (lower 95% limit, upper 95% limit, mm) were (A) = (-0.013, 0.014), (S)=(-0.004, 0.015), (L/R)=(-0.070, 0.050). For the human subject measurements, the mean RC (mm) per plate were (A)=0.397, (S)=0.354, (L/R)=0.743. The ICC were (A)=1.000, (S)=1.000, (L/R)=1.000. The mean LOA (lower limit, upper limit, mm) were (A)=(-0.314, 0.487), (S)=(-0.194, 0.341), (L/R)=(0.027, 0.506).
     This study did not compare different measurement methods.
     The study demonstrates sufficient (<1mm) intraobserver repeatability and interobserver agreement for both phantom and human subjects for use in a multi-fraction Gamma Knife setting. Phantom measurements showed better repeatability than human measurements. Left/right plate measurements may be slightly less reliable than anterior and superior plates.
     The results suggest it is not necessary to ensure a single observer takes the measurements across all fractions of an Extend treatment.


Project Roles:

D. Schlesinger (), Z. Xu (), C. Yen (), J. Sheehan ()