A preliminary study on quantitative assessment of apical lesion area

To establish a quantitative method for calculating the area of apical lesions, thereby enabling a more precise assessment of the extent of apical lesions.

Three observers annotated periapical radiographs using a proprietary web-based annotation platform. Root outlines and lesion boundaries were delineated to obtain both the absolute lesion area (S_a) and the relative lesion area (S_r), calculated as the ratio of lesion area to root area. Observers also evaluated periapical status using the Periapical Index (PAI). All measurements were repeated after one month. Intra- and inter-observer reliability for PAI scores were assessed using Cohen's kappa and Fleiss' kappa, respectively, while reliability for S_a and S_r measurements was evaluated using the intraclass correlation coefficient (ICC). Agreement between repeated measurements was visualized using Bland-Altman plots.

Annotations were completed on 126 radiographs containing 490 tooth roots. PAI scores demonstrated moderate intra-observer (Cohen's kappa range: 0.45-0.58) and inter-observer (Fleiss' kappa: 0.41) reliability. In contrast, ICC values for both S_a and S_r indicated good to excellent reliability (range: 0.72-0.89). Intra-observer ICCs for S_r (range: 0.85-0.89) were consistently higher than those for S_a (range: 0.72-0.81). Bland-Altman analysis revealed that over 95% of the measurement differences fell within the limits of agreement for all observers.

This study established a relatively quantitative method for assessing apical lesions. S_r demonstrated the highest consistency, followed by S_a, while PAI showed only moderate reliability. These findings support the potential of quantitative metrics as objective alternatives to traditional subjective scoring systems.