Table 6.
Values of positive and negative likelihood ratios (LR), subdivided by image method, with relative weighted average
| Imaging method | References | Patients n | LR+ (% patients) | LR− (% patients) | PPV (%) | NPV (%) | Weighted average | Advantages | Disadvantages |
|---|---|---|---|---|---|---|---|---|---|
| CT | Handschel et al. [15] | 107 | 6.30 (44) | 0.20 (44) | 82.61 | 86.89 | LR (T+)=14.80 | Medium high specificity | -The thickness of the scan can influence the sensitivity.15 |
| Gu et al. [14] | 46 | 42.10 (19) | 0.58 (19) | 100.00 | 82.93 | LR (T−)=0.35 | -Possible artifacts generated by metal rehabilitations14 | ||
| Van Cann et al. [9] | 66 | 13.50 (27) | 0.43 (27) | 96.15 | 55.00 | PPV=90.31 | -Underestimate the extent of the bone invasion21 | ||
| Van den Brekel et al. [13] | 23 | 3.31 (9.5) | 0.45 (9.5) | 90.00 | 61.54 | NPV=74.91 | High specificity | -In T1–T2 windows with the use of gadolinium, overestimation of the tumor extension is possible if edema is present.13 | |
| MRI | Gu et al. [14] | 46 | 20.10 (21) | 0.42 (21) | 87.50 | 86.34 | LR (T+)=37.90 | ||
| Hendrikx et al. [8] | 23 | 91.80 (11) | 0.16 (11) | 69.23 | 80.00 | LR (T−)=0.24 | |||
| Van Cann et al. [9] | 66 | 63.40 (31) | 0.37 (31) | 100.00 | 58.97 | PPV=90.63 | -Possible artifacts due to movements, periodontal inflammations and partial volume defects14 | ||
| Van den Brekel et al. [13] | 29 | 3.51 (13.5) | 0.07 (13.5) | 85.00 | 88.89 | NPV=78.69 | |||
| Kolk et al. [16] | 50 | 15.80 (23) | 0.05 (23) | 94.44 | 91.67 | -Signal weakness for cortical bone16 | |||
| CBCT | Hendrikx et al. [8] | 23 | 91.80 (28) | 0.09 (28) | 100.00 | 92.31 | LR (T+)=27.80 | -High sensitivity and specificity. | -Underestimates the extent of bone invasion by the tumor8 |
| Hakim et al. [12] | 58 | 2.44 (71) | 0.11 (71) | 72.50 | 88.89 | LR (T−)=0.10 | -It requires a lower radiation load than the CT and the MSCT8 | -Weak contrast of soft tissues12 | |
| PPV=80.05 | |||||||||
| NPV=89.83 | -Reduced soft tissue distortion due to gravity.8 | -It is altered by inflammatory processes or an increase in hematopoiesis14 | |||||||
| SPECT | Hakim et al. [12] | 62 | 1.85 (76) | 0.08 (76) | 64.44 | 94.12 | LR (T+)=25.50 | -High sensitivity-Identifies hyper-hypometabolic and hypometabolic lesions14 | |
| Kolk et al. [16] | 20 | 99.00 (24) | 0.009 (24) | 100.00 | 100.00 | LR (T−)=0.06 | |||
| PPV=72.97 | |||||||||
| NPV=95.53 | |||||||||
| MSCT | Hakim et al. [12] | 78 | 3.31 (61) | 0.45 (61) | 75.00 | 74.00 | LR (T+)=37.00 | High specificity | -Tendency to underestimate the extent of the tumor8 |
| Kolk et al. [16] | 50 | 89.80 (39) | 0.11 (39) | 100.00 | 73.33 | LR (T−)=0.31 | -Possible artifacts generated by metal rehabilitations9 | ||
| PPV=87.44 | |||||||||
| NPV=73.74 | |||||||||
| PR | Hendrikx et al. [8] | 23 | 6.56 (31) | 0.49 (31) | 85.71 | 68.75 | LR (T+)=4.88 | -Low dosimetry | -Low sensitivity in detecting cortical bone erosion13 |
| Kolk et al. [16] | 50 | 4.11 (68) | 0.31 (68) | 82.35 | 69.23 | LR (T−)=0.36 | -Use to detect periodontal and periapical lesions of not clear interpretation with other modalities9 | ||
| PPV=84.24 | |||||||||
| NPV=69.18 |
LR+: positive likelihood ratio; LR−: negative likelihood ratio; PPV: positive predictive value; NPV: negative predictive value; T+: test positive; T−: test negative