Performance of the Malmgren Index for Assessing Root Resorption on 2D vs. 3D Radiographs: A Pilot Study

Objectives: To compare the performance of the Malmgren index on 2D and 3D radiographs.

Methods: Patients with a panoramic radiograph and a cone beam computed tomography (CBCT) taken at an interval of <3 months and presenting root resorption (RR) on at least one incisor and/or canine were retrospectively included. RR was scored twice by two observers using the Malmgren index in both the 2D and 3D sets, and intra-class correlation coefficient (ICC) was calculated.

Results: 155 teeth were analyzed. The ICC was the lowest in 2D, followed by overall, transversal and sagittal 3D. Malmgren scores were systematically higher in 2D, which overestimated RR, especially in the transversal plane on all incisors and canines and in the sagittal plane on the maxillary incisors. 2D respectively leads to 28.0-34.8% of false positives and negatives when discriminating between RR or not. The early stages of RR are often misdiagnosed in 2D, while later stages are more accurate.

Conclusions: The original Malmgren index is not suited for 3D images, especially axial, where using dichotomized values (resorption yes/no) leads to overestimation of RR. A low-dose CBCT of the upper incisors could detect RR with high diagnostic accuracy in the early stages of orthodontic treatment, especially in patients with dental trauma or familial RR history.