Table 2.
Relationship between NICE classification and pathological findings, and diagnostic performance of NBI, Laser-BLI and LED-BLI
| Pathological findings | Diagnostic performance | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| NICE classification | N (%) | HP/SSL | LGD+HGD+s-SM | d-SM | Sensitivity | Specificity | PPV | NPV | Accuracy | Cronbach’s alpha | |
| NBI | Type 1 | 176 (23.0) | 125 | 51 | 0 | 85.6 (79.9-91.3) | 91.8 (89.6-93.9) | 71.0 (64.3-77.7) | 96.4 (94.9-97.9) | 90.6 (88.5-92.7) | |
| Type 2 | 583 (76.2) | 21 | 562 | 0 | 91.4 (89.2-93.6) | 88.4(83.2-93.6) | 96.4 (94.9-97.9) | 70.1(63.5-76.7) | 90.3 (88.2-92.4) | ||
| Type 3 | 6 (0.8) | 0 | 2 | 4 | 100 (100.0-100.0) | 99.7 (99.4-100.1) | 66.7 (28.9-104.4) | 100 (100.0-100.0) | 99.5 (99.0-100.0) | ||
| Total | 765 (100) | 146 | 615 | 4 | 0.961 | ||||||
|
Laser- BLI |
Type 1 | 231 (26.5) | 210 | 21 | 0 | 82.7 (78.0-87.3) | 96.6 (95.2-98.0) | 90.9 (87.2-94.6) | 93.1 (91.2-95.1) | 92.5 (90.8-94.3) | |
| Type 2 | 638 (73.2) | 44 | 594 | 0 | 96.1 (94.6-97.6) | 82.7 (78.0-87.3) | 93.1 (91.1-95.1) | 89.7 (85.9-93.6) | 92.2 (90.4-94.0) | ||
| Type 3 | 3 (0.3) | 0 | 3 | 0 | N/A | 99.7 (99.3-100.0) | 0 (0.0-0.0) | 100 (100.0-100.0) | 99.7 (99.3-100.0) | ||
| Total | 872 (100) | 254 | 618 | 0 | 0.962 | ||||||
|
LED- BLI |
Type 1 | 231 (31.0) | 179 | 57 | 0 | 84.0 (79.1-89.0) | 89.6 (87.0-92.2) | 75.8 (70.4-81.3) | 93.5 (91.4-95.6) | 88.0 (85.7-90.3) | |
| Type 2 | 525 (69.0) | 34 | 491 | 0 | 89.6 (87.0-92.2) | 84.0 (79.1-89.0) | 93.5 (91.4-95.6) | 75.8 (70.4-81.3) | 88.0 (85.7-90.3) | ||
| Type 3 | 0 (0) | 0 | 0 | 0 | N/A | 100 (100.0-100.0) | N/A | 100 (100.0-100.0) | 100 (100.0-100.0) | ||
| Total | 761 (100) | 213 | 548 | 0 | 0.917 | ||||||
This table shows cumulative results of the four evaluators
NBI narrow band imaging, Laser-BLI blue laser imaging, LED-BLI blue light imaging, NICE NBI international colorectal endoscopic, SSL sessile serrated lesion, s-SM shallow submucosal, d-SM deep submucosal, PPV positive predictive value, NPV negative predictive value