Comparison of diagnostic efficiency between fluorescence
colposcope and conventional colposcopy
The comparison results of colposcopy and pathological examination were
illustrated in Table 2. We analyzed the diagnostic efficiency of
fluorescence colposcopy
on a per-patient basis and
per-site basis separately. On the per-patient basis, the
ROC curve of each patient’s SBRs
was shown in Figure 3A. The set
cut-off value was 2.34,
correspondingly, there were 10, 32, 50, and 5 patients with true
positive (TP), false positive (FP), true negative (TN), and false
negative (FN) respectively (Figure
3B, Table 2). A total of 360
cervical sites were sampled in the 97 patients of fluorescence
colposcope group. The ROC curve of SBRs on the per cervical site basis
was displayed in Figure 3C and the cut-off value was 1.63. The cervical
sites of TP, FP, TN, and FN were
16, 87, 250, 7 respectively (Figure 3D, Table 2).
For conventional colposcope,
comparing the colposcopic impressions with the pathological results, the
number of TP, FP, TN, and FN cases were 13, 29, 45, 11 respectively
(Table 2).
The diagnostic efficiency
including accuracy, sensitivity,
specificity, PPV, NPV and associated 95% CI were calculated and shown
in Table 3. The diagnostic
efficiency of fluorescence
colposcope on a per-patient basis
(accuracy=61.9%,
sensitivity=66.7%, specificity=61.0%, PPV=23.8%, NPV=90.9%) was
similar to the conventional
colposcope group (accuracy=59.2%,
sensitivity=54.2%,
specificity=60.8%, PPV=31.0%, NPV=80.4%). However, when performing
statistical analysis on a per-site
basis, the accuracy, specificity, and NPV of the fluorescence colposcopy
group (73.9%, 74.2%, 97.3%
respectively) were better than
those of the conventional colposcopy group (p=0.005, 0.021, 0.000
respectively), but the PPV (15.5%) was lower (p=0.035). These results
indicated that fluorescence colposcope had a comparable effect in
diagnosing cervical precancerous lesions compared with conventional
colposcope. Surprisingly, this fluorescence colposcope also demonstrated
an excellent negative predictive effect, which was valuable for cervical
cancer screening.