Table 7.

Diagnostic accuracy of PR and PAN in the detection of AP lesions and methods used to calculate them in the studies without a direct index

Study	Description of diagnostic accuracy measurement
Estrela et al., 2008 [50]	Diagnostic accuracy was 70.0% for PR and 54.0% for PAN. Sensitivity, specificity, PPV and NPV were 55.0%, 98.0%, 98.0% and 55.0% for PR and 28.0%, 100%, 99.0% and 44.0% for PAN, respectively
Estrela et al., 2009 [43]	In a sample of 1,020 analyzed teeth, AP lesions were detected in 397 teeth (38.9%) using PR and in 614 teeth (60.2%) using CBCT. The diagnostic accuracy was calculated with the ratio 397:614 = X:100, where "X" represented the diagnostic accuracy as a percentage of the PR compared to CBCT used as a reference standard. Diagnostic accuracy was 64.7% for PR
Moura et al., 2009 [44]	Percentage values ​​regarding AP lesions were provided in two different root canal obturation situations: 1–2 mm from the apex and 1–2 mm beyond the apex. The study analyzed three anatomical sites (anterior teeth, premolars and molars), using both PR and CBCT. To calculate diagnostic accuracy, averages were made between the percentages of the two-root canal obturation lengths for each anatomical area. This was done to avoid making distinctions between the two-root canal obturation lengths, as it was not relevant to the systematic review in question. From the calculated averages, it emerged that AP lesions were found on PR in 16.5%, 11.7% and 27.3% in the anterior, premolar and molar areas, respectively. In CBCT, after calculating the averages in the same way, the percentage values ​​were 24.5% in the anterior teeth, 14.5% in premolars and 36.3% in molars. Proportions were calculated to determine the diagnostic accuracy between PR and CBCT for each anatomical area, using CBCT as a reference standard. For example, for the anterior teeth, the ratio was 16.5: 24.5 = X:100, where "X" represents the diagnostic accuracy of PR. After calculating the three proportions, one for each anatomical area, it emerged that the diagnostic accuracy in anterior teeth was 67.3%, in premolars 80.3% and in molars 75.2%. Finally, the percentage accuracy values ​​were averaged to report an overall value. At the end, diagnostic accuracy was 74.3% for PR
Weissman et al., 2015[45]	It was reported that the probability of finding AP lesions was 56.7% with PR and 79.1% with CBCT. Since CBCT was considered the reference standard in this study, a proportion was made: 56.7:79.1 = X:100. Diagnostic accuracy was 71.7% for PR. To calculate the other parameters, AP lesions detected both by PR and CBCT were considered true positives (38) while those not detected on PR but visible on CBCT were considered false negatives (15). False positives were AP lesions detected on PR but not on CBCT (0) and true negatives when neither PR or CBCT detected any lesion (14). On this basis, sensitivity, specificity, PPV and NPV were 72.0%, 100%, 100% and 48.0%
Kanagasingam et al., 2017 [39]	Area under the curve (AUC) was provided as a measure of diagnostic accuracy for several imaging methods: FP (single intraoral film), FPS (3-projection intraoral film: normal, disto and mesio angled 10°), DP (single intraoral digital view), DPS (digital intraoral in the 3 projections). The diagnostic accuracy for the FP method was found to be 56.2%, for the FPS method 68.5%, for the DP method 62.9% and for the DPS method 68.8%. After averaging these values, an overall total diagnostic accuracy was 64.1% for PR using histopathological examination as the reference standard. Sensitivity, specificity, PPV and NPV were, again after averaging, 29.5%, 99.5%, 99.5% and 40.5%
Kanagasingam et al., 2017 [40]	Area under the curve was provided but only AP lesions in digital view and CBCT was considered. AUC value for DP method was 62.9%, for DPS method 68.8% and for CBCT 94.3%, using histopathological examination as a reference standard. An average was calculated between the two accuracy values ​​for the digital PR in the two methods. Diagnostic accuracy was 65.85% for PR. Sensitivity, specificity, PPV and NPV were reported for DP and DPS, and after averaging, the overall values were 32.5%, 99.0%, 99.0% and 41.5%
Nardi et al., 2017 [47]	Diagnostic accuracy was 65.0% for PAN. Sensitivity, specificity, PPV and NPV were 34.2%, 95.8%, 89.1% and 59.3%
Nardi et al., 2018 [48]	Diagnostic accuracy was 71.3% for PAN. Sensitivity, specificity, PPV and NPV were 48.8%, 93.8%, 88.6% and 64.7%
Nardi et al., 2020 [49]	Diagnostic accuracy values ​​of PAN in the detection of AP lesions were reported, both in treated and non-endodontically treated teeth. Diagnostic accuracy for endodontically treated teeth was 73.4%, while it was 67.6% for non-endodontically treated teeth. Average diagnostic accuracy was 70.0% for PAN. Likewise, average sensitivity, specificity, PPV and NPV were 45.9%, 96.3%, 92.4% and 64.2%
Jang et al., 2020 [46]	AP lesions were detected in 24.6% of cases using PR and in 35.5% of cases using CBCT. Since CBCT was considered the reference standard in this study, a proportion was calculated using CBCT as the reference, which was 24.6:35.5 = X:100. As result diagnostic accuracy was 69.4% for PR
Bosoni et al., 2021 [42]	Diagnostic accuracy was 72.1% for PAN. Sensitivity, specificity, PPV and NPV were 46.8%, 97.4%, 94.7% and 64.7%
Li et al., 2022 [41]	Data regarding the diagnosis of AP lesions using an AI-based method proposed in the same study were available. Using clinically diagnosed PR as a reference standard, the diagnostic accuracy of the AI ​​method was calculated. By adding the true positives and true negatives and dividing the result by the total number of cases—true positives and true negatives and false positives and false negatives—diagnostic accuracy was 83.0% for PR by AI. Sensitivity, specificity, PPV and NPV were 82.0%, 84.0%, 83.7% and 82.4%