Purpose Assessment of periodontal bone resorption severity from panoramic radiographs can be time-consum-

ing and subject to inter-observer variability. This study evaluates a multimodal deep learning framework for auto-

mated Bone Resorption Age Ratio (BRAR)-based severity classification by integrating panoramic radiographic

features with structured metadata.

Methods Using 988 annotated panoramic radiographs from a publicly available dataset, image-only, meta-

data-only, and multimodal models were evaluated across 20 repeated stratified random splits. The multimodal

model integrates DenseNet-121 radiographic features with a multilayer perceptron encoding structured meta-

data through feature-level concatenation and progressive fine-tuning. Performance was assessed using accuracy,

macro F1-score, macro area under the receiver operating characteristic curve (macro AUC-ROC), average preci-

sion, and class-specific metrics. Paired comparisons of three primary outcomes (macro F1, macro AUC-ROC,

and Level 3 recall) used the Wilcoxon signed-rank test with Holm–Bonferroni correction. A four-way metadata

ablation assessing the contributions of age and the remaining structured variables, an exploratory probability-

level ensemble, qualitative Gradient-weighted Class Activation Mapping (GradCAM), and a targeted expert error

analysis on 88 purposively selected cases were also performed.

Results The multimodal model achieved an accuracy of 0.67 ± 0.03, macro F1-score of 0.65 ± 0.03, macro AUC-

ROC of 0.81 ± 0.03, and Level 3 recall of 0.60 ± 0.07. It outperformed the image-only model in macro F1 and macro

AUC-ROC (both Holm-adjusted ???????? ???? ????????????????????) and Level 3 recall (???????? ???? ????????????????????), with all three primary outcomes also

significantly higher than the metadata-only model (???????? ???? ????????????????????), which identified only 0.14 ± 0.07 of severe cases. In

an ablation of the metadata inputs, neither age alone nor the five non-age variables alone significantly improved

macro F1 or Level 3 recall over the image-only model, whereas the full model significantly outperformed both. In

the targeted expert subset, dentist–original agreement was higher among model-correct cases (???????? ???? ????????????????????; weighted

???????? ???? ????????????????????) than among model-misclassified cases (???????? ???? ????????????????????; weighted ???????? ???? ????????????????????).

Conclusions Integrating panoramic radiographs with structured metadata improved prediction of BRAR-derived

periodontal severity labels compared with either modality alone. The metadata ablation showed that age alone

did not account for the multimodal improvement, with the findings supporting complementary contributions

from age and the remaining structured metadata. The targeted expert analysis showed greater dentist–annotation 

disagreement among selected model-misclassified cases but, because it involved one dentist and a purposively

selected subset, does not establish annotation uncertainty as the cause of model errors. External validation, auto-

mated localization, and evaluation of direct radiographic bone-resorption estimation followed by age normalization are important directions for future work.

 

Title

BMC Medical Imaging

Publisher

BioMed central BMC Springer Nature

Publisher Country

Palestine

Publication Type

Prtinted only

Volume

26

Year

2026

Pages

--