• 2022-09
  • 2022-08
  • 2022-07
  • 2022-06
  • 2022-05
  • 2022-04
  • 2021-03
  • 2020-08
  • 2020-07
  • 2020-03
  • 2019-11
  • 2019-10
  • 2019-09
  • 2019-08
  • 2019-07
  • CP-456773 br this is standard clinical practice In the case


    this is standard clinical practice. In the case of CBCT-US registration (henceforth referred to as CBCT-US fusion), the planning CT was available for reference in a separate window rather than as a third superimposed image.
    Image contouring and rating
    Observers used the MATLAB application for all image analysis, including
    1. Delineation of the uterus on a pr-selected 2D sagittal slice: The sagittal slice used for contouring was the centermost slice of the uterus, as identified by one of the observers in the Experienced cohort (described in next section). This slice was in the same position for corre-sponding CBCT, US, and CBCT-US fusion images.
    2. Contour confidence ratings: After contouring the uterus, each observer was asked to rate the confidence that his or her contour reflected the true uterine boundary on a scale from 1 (extremely unconfident) to 10 (extremely confident).
    Observers first evaluated CBCT and US images sepa-rately. These images were displayed in a random order so that observers were blind to which CBCT and US images were paired and which images were from the same patient. Observers then evaluated the CBCT-US fusion images, which were also displayed in a random order.
    One experienced observer also contoured the CP-456773 on the central 2-dimensional sagittal slice on US and CBCT so that the relationship between bladder size and image quality could be assessed.
    Observer selection and training
    Eight observers assessed US, CBCT, and CBCT-US fusion image quality in this study. Observers were divided into 2 cohorts:
    1. Experienced (3 observers): One clinician (IW) and 2 medical physicists (TOS and SM) with previous expe-rience in interpreting both US and CBCT images.
    2. New-to-US (5 observers): Clinicians, radiographers, and medical physicists who may have had experience in CBCT image analysis but had no prior training in interpreting US images.
    Before performing the analysis, all 8 observers partici-pated in a 1-hour training session (using the training dataset composed of 17 CBCT, US, and CBCT-US fusion images) designed to (1) teach observers how to use the MATLAB software application; (2) give observers experience in interpreting and contouring US, CBCT, and CBCT-US fusion images of the uterus; and (3) establish a consensus for rating contour confidence.
    688 Mason et al. International Journal of Radiation Oncology Biology Physics
    Contour agreement metrics
    The two geometric measures used to assess contour agreement were the Dice similarity coefficient (DSC)27 and the mean contour-to-contour distance (MCCD). For 2 contours X and Y, the DSC was calculated as (2jXXYj)/ (jXjþjY j), with 0 and 1 representing no overlap and perfect overlap, respectively, and the MCCD was defined as 1 n
    where n is the number of points comprising contour X.28
    Generation of the gold standard contour
    For every image analyzed, a single contour was generated by combining the 3 contours from each of the observers in the Experienced cohort using Simultaneous Truth and Performance Level Estimation (STAPLE).29 The gold standard contour generated by the STAPLE method is the contour that optimizes the sensitivity and specificity of all of the contours from each individual observer. Because 40 CBCT, 40 US, and 40 CBCT-US fusion images were analyzed, this resulted in the generation of 120 gold stan-dard contours. The contour agreement among the remaining 5 contours from the New-to-US cohort and the gold stan-dard STAPLE contour was measured for each of the 120 images.
    Observer ratings
    Observers were required to rate how confident they were that their contours reflected the true uterine boundary on a 10-point scale. Observers were asked to consider all factors potentially influencing their rating, including the visibility of the uterine boundary, the clarity of the bladder wall and bowel gas, the regularity of the uterine shape, and the po-sition of anatomic landmarks with respect to the planning target volume. All observers had practice in using these factors to inform their rating during the 1-hour training session.