Robust Estimation of the Chronological Age of Children and Adolescents Using Tooth Geometry Indicators and POD-GP

Abstract

Determining the chronological age of children or adolescents is becoming an extremely necessary and important issue. Correct age-assessment methods are especially important in the process of international adoption and in the case of immigrants without valid documents confirming their identity. It is well known that traditional, analog methods widely used in clinical evaluation are burdened with a high error rate and are characterized by low accuracy. On the other hand, new digital approaches appear in medicine more and more often, which allow the increase of the accuracy of these estimates, and thus equip doctors with a tool for reliable estimation of the chronological age of children and adolescents. In this study, the work on a fast and effective metamodel is continued. Metamodels have one great advantage over all other analog and quasidigital methods-if they are well trained, a priori, on a representative set of samples, then in the age-assessment phase, results are obtained in a fraction of a second and with little error (reduced to ±7.5 months). In the here-proposed method, the standard deviation for each estimate is additionally obtained, which allows the assessment of the certainty of each result. In this study, 619 pantomographic photos of 619 patients (296 girls and 323 boys) of different ages were used. In the numerical procedure, on the other hand, a metamodel based on the Proper Orthogonal Decomposition (POD) and Gaussian processes (GP) were utilized. The accuracy of the trained model was up to 95%.

Keywords: Gaussian processes; age assessment; artificial intelligence; chronological age; dental age; digital image analysis; digital pantomography; proper orthogonal decomposition.

Figure 1

Graphical representation of indicator X01 (red |C13C43|; blue |C15C45|).

Figure 2

Graphical representation of selected indicators: X01 (red |C13C43|; blue |C15C45|), X02 (red |C13C43|; green |C16C46|), X03 (red |C13C43|; orange |C17C47|).

Figure 3

Mean absolute error of training and testing data computed by a model without POD truncation and with different number of testing and training sets.

Figure 4

The results of sensitivity analysis for all 21 indicators.

Figure 5

SAME for testing and training set with respect to number of input amplitudes.

Figure 6

Differences between the values predicted by the 7-amplitude model and the actual values.

Figure 7

Differences between the values predicted by the 7-amplitude model and the actual values with respect to reference values (targets).

Figure 8

R square value for training and testing sets generated by different models: (a) training set—model with 7-amplitudes; (b) testing set—model with 7 amplitudes; (c) training set—model with 17 amplitudes; (d) testing set—model with 17 amplitudes.

Figure 9

Patient age distribution.

Figure 10

Statistical properties of the indicators X01–X21.