From: Fully automated method for dental age estimation using the ACF detector and deep learning
Authors | Input | Automated type | Data | Stage assessment | Age estimation performance (difference and correlation between estimated age and chronological age) | ||
---|---|---|---|---|---|---|---|
Age range of samples (years) | No. of images | Demirjian’s method | Accuracy | ||||
Zaborowicz et.al., (2022) | Extracted tooth and bone parameters using ImageJ software | Semi | 4–15 | 619 | - | - | Difference: median absolute error = 1.48 years and mean absolute error = 4.61 months Correlation: R2 = 0.93 |
Vila-Blanco et.al., (2020) | Full image | Fully | 4.5–89.2 | 2289 | - | - | Difference: median absolute error = 1.48 years and mean absolute error = 2.84 years Correlation: R2 = 0.90 |
Hou et.al., (2021) | Full image | Fully | 0–90 | 27,957 | - | - | Difference: mean absolute error = 1.64 years |
Atas et.al., (2022) | Full image | Fully | 8–68 | 1332 | - | - | Difference: mean absolute error = 3.13 years Correlation: R2 = 0.87 |
Milošević et.al., (2022) | Specific teeth image | Full images: fully Individual teeth: semi | 19–90 | 4035 | - | - | Difference: median absolute error = 2.95 years for full images, median absolute error = 4.68 years for individual teeth |
De Tobel et.al., (2017) | Specific teeth image | Semi | 7–24 | 400 | 10 stages (modified) | 51.00% | - |
Merdietio Boedi et.al., (2020) | Specific teeth image | Semi | 7–24 | 400 | 10 stages (modified) | 61.00% | - |
Banar et.al., (2020) | Specific teeth image | Fully | 7–24 | 400 | 10 stages (modified) | 54.00% | - |
Upalananda et.al., (2021) | Specific teeth image | Semi | 15–23 | 2235 | 5 stages (D to H) | 82.50% | Correlation: Spearman’s rank correlation r = 0.87, p < 0.001 |
Proposed method | Specific teeth image | Fully | 15–23 | 1000 | 5 stages (D to H) | 83.25% | Difference: median absolute error = 1.72 years and mean absolute error = 1.94 years Correlation: Pearson’s linear correlation ρ = 0.91, p < 0.05 |