Table 2.
Cutoff threshold and corresponding sensitivity, specificity, NPV and PPV for detecting RL using FDG PET. In total, there were 20 RL and 58 NRL
| Index | Threshold (number rated NRL-number rated RL) | Sensitivity | Specificity | NPV | PPV | Accuracy |
|---|---|---|---|---|---|---|
| Using cutoff threshold of 15 % | ||||||
| ΔSUVmax_BW | 15 % (35-43) | 80 % | 53 % | 89 % | 37 % | 60 % |
| ΔSUV40 %_BW | 15 % (32-46) | 95 % | 53 % | 97 % | 41 % | 64 % |
| ΔSUVmax_BSA | 15 % (35-43) | 80 % | 53 % | 89 % | 37 % | 60 % |
| ΔSUV40 %_BSA | 15 % (33-45) | 95 % | 55 % | 97 % | 42 % | 65 % |
| Using cutoff threshold of 25 % | ||||||
| ΔSUVmax_BW | 25 % (46-32) | 65 % | 67 % | 85 % | 41 % | 67 % |
| ΔSUV40 %_BW | 25 % (39-39) | 90 % | 64 % | 95 % | 46 % | 71 % |
| ΔSUVmax_BSA | 25 % (46-32) | 65 % | 67 % | 85 % | 41 % | 67 % |
| ΔSUV40 %_BSA | 25 % (39-39) | 90 % | 64 % | 95 % | 46 % | 71 % |
| Aiming at a 95 % sensitivity of detection of RL | ||||||
| ΔSUVmax_BW | 14 % (32-46) | 95 % | 53 % | 97 % | 41 % | 64 % |
| ΔSUV40 %_BW | 22 % (38-40) | 95 % | 64 % | 97 % | 47 % | 72 % |
| diffSUVmax_BW | 1.4 (33-45) | 95 % | 55 % | 97 % | 42 % | 65 % |
| diffSUV40 %_BW | 1.2 (41-37) | 95 % | 69 % | 98 % | 51 % | 76 % |
| Aiming at a 95 % NPV for detection of RL | ||||||
| ΔSUVmax_BW | 14 % (32-46) | 95 % | 53 % | 97 % | 41 % | 64 % |
| ΔSUV40 %_BW | 22 % (38-40) | 95 % | 64 % | 97 % | 47 % | 72 % |
| diffSUVmax_BW | 1.4 (33-45) | 95 % | 55 % | 97 % | 42 % | 65 % |
| diffSUV40 %_BW | 1.2 (41-37) | 95 % | 69 % | 98 % | 51 % | 76 % |