Image Quality; Table 13 Statistics And 2X2 Cross-Tabulation Components By Grader - Clarity RetCam3 User Manual

Table 12c: Cross-classification from Grader C versus BIO (35-37 Weeks PMA)
Type 2 pre-threshold ROP or
treatment-requiring ROP
Unknown
No ROP or Mild ROP
Total
Based on a review of the RetCam images, Grader C concluded that 40 sets of images had
Type 2 pre-threshold ROP or Treatment-requiring ROP including all 26 cases determined
by BIO to be Type 2 pre-threshold ROP or Treatment-requiring ROP. Grader C was able to
make a determination for all of the sets of images. Eighty (80) sets of images were
concluded by Grader C to have no ROP or mild ROP.

Table 13 Statistics and 2x2 Cross-Tabulation Components by Grader

Table 13: Performance for Type 2 Threshold ROP or Worse
Grader
A
B
C
* This counts as false positives the 6 sets of images classified as unknown by Grader A
and 'mild ROP or less' by the reference standard.
** This counts as false positives the 8 sets of images classified as unknown by Grader B
and 'mild ROP or less' by the reference standard.
Unknowns in this study were treated as false positives (referral-warranted disease) for
statistical purposes in
examination by BIO. None of the BIO referrals were missed by the RetCam Graders
resulting in 100% sensitivity results for each Grader. Due to "over referral" by two (2) of the
Graders and the inclusion of unknowns as referrals, specificity results for the graders
ranged from 85% to 88%.

Image Quality

Before providing a diagnosis for each image set, the three (3) retinal specialist graders
assessed the technical quality of the images for "adequate", "possibly adequate", or
"inadequate". Each of them found that images taken at 35-37 weeks PMA by trained NICU
nurses were technically "adequate" or "possibly adequate" at a rate of 93.3% to 100%.
RetCam 3 and RetCam Shuttle User Manual
Screening for Retinopathy of Prematurity (ROP)
Grader C
Type 2 pre-threshold ROP or
treatment-requiring ROP
26
0
0
26
Sensitivity ~95% CI for
Sensitivity
1.0 (26/26) (0.868, 1.0)
1.0 (26/26) (0.868, 1.0)
1.0 (26/26) (0.868, 1.0)
Table 13 because in the clinical setting such a result would result in
Reference Standard: BIO
No ROP or Mild ROP
14
0
80
94
Specificity ~95% CI for
Specificity
0.883 (83/94)* (0.802, 0.933)
0.851 (80/94)** (0.765, 0.909)
0.851 (80/94) (0.765, 0.909)
147
PN 18-000393 Rev. A