MATERIALS AND METHODS: This parallel, single-blinded, randomised controlled trial (RCT) consisted of 22 periodontitis patients who had molar with advanced furcation involvement (FI). All patients followed the same inclusion criteria and were treated following the same protocol, except for radiographic evaluation (CBCT vs. periapical). This study proposed and evaluated five parameters that represent the extent and severity of furcation defects in molars teeth, including CEJ-BD (clinical attachment loss), BL-H (depth), BL-V (height), RT (root trunk), and FW (width).

Method: Maxillary CBCT images of two-hundred-and-fifty-seven consecutive patients (163 men, 94 women, mean age 42 years) were analyzed. Samples were later divided into dentate (n = 142) and posteriorly edentulous (n = 115) jaws. Using both alveolar ridge and tooth location as reference points, the distance and diameter of IA were assessed.

Result: The IA was seen in 63.7% of all sinuses with 68.2% in dentate and 62.4% in edentulous. Mean distance and diameter of IA across the posterior tooth locations were 17.9 ± 3.0 mm and 1.4 ± 0.5 mm (dentate) and 15.1 ± 3.0 mm and 1.0 ± 0.5 mm (posteriorly edentulous), respectively. In each sample, there were no significant differences in distance-alveolar ridge and no significant correlations in diameter-tooth location. A statistically significant Pearson coefficient correlation between diameter and distance in dentate state was observed (r = -0.6).

METHODS: External root resorption was simulated on 88 extracted premolar teeth using tungsten bur in different depths (0.5 mm, 1 mm, and 2 mm). All teeth were scanned using a Cone beam CT (Carestream Dental, Atlanta, GA). Afterward, a training (70%), validation (10%), and test (20%) dataset were established. The performance of four DLMs including Random Forest (RF) + Visual Geometry Group 16 (VGG), RF + EfficienNetB4 (EFNET), Support Vector Machine (SVM) + VGG, and SVM + EFNET) and four hybrid models (DLM + FST: (i) FS + RF + VGG, (ii) FS + RF + EFNET, (iii) FS + SVM + VGG and (iv) FS + SVM + EFNET) was compared. Five performance parameters were assessed: classification accuracy, F1-score, precision, specificity, and error rate. FST algorithms (Boruta and Recursive Feature Selection) were combined with the DLMs to assess their performance.

RESULTS: RF + VGG exhibited the highest performance in identifying ERR, followed by the other tested models. Similarly, FST combined with RF + VGG outperformed other models with classification accuracy, F1-score, precision, and specificity of 81.9%, weighted accuracy of 83%, and area under the curve (AUC) of 96%. Kruskal Wallis test revealed a significant difference (p = 0.008) in the prediction accuracy among the eight DLMs.