## Evaluation of the performance of caries detection methods

The performance of caries detection methods should be assessed considering two important parameters: reproducibility and validity. A reproducible method is the one that presents similar results and shows an agreement between two exams performed in different moments or by different examiners using the same sample. Reproducibility can be assessed by Cohen's Kappa test or Intraclass Correlation Coefficient (ICC).

Validity is the ability of a method of assessing what it is suppose to assess. It is calculated by the proportion of correct results taking into account the gold standard, which is the true and definitive diagnosis reference. Using these results, the validity of a method can be obtained by calculating values of specificity and sensitivity. Specificity is the proportion of cases classified by a method as negative (disease absent) considering the total of cases that did not developed the disease. Sensitivity, however, is the proportion of cases classified as positive (disease present) considering the total of cases that really developed the disease. The total percentage of correctly assessed cases considering the presence and the absence of disease is represented by the accuracy. Table 1 summarizes how sensitivity (Sn) and specificity (Sp) values can be calculated:

 Disease present (+) Disease absent (-) Positive Test (+) A B Total positive tests (A + B) Negative Test (-) C D Total negative tests (C + D) Sn%: A/(A + C) Sp%: D/(B + D) Total number (A + B + C + D)

Table 1. The generic 2 x 2 table used to calculate sensitivity (Sn) and specificity (Sp) values.

Table 1. The generic 2 x 2 table used to calculate sensitivity (Sn) and specificity (Sp) values.

Concerning methods' validity and calculation of sensitivity and specificity values it is necessary to establish limits to define what "disease" and "healthy" mean considering the gold standard. These limits can also be called "cut-off points", which are combined according to the criteria used for the gold standard classification. For example, caries lesions can be classified in: (0) caries free, (1) caries extending up to halfway through the enamel, (2) caries extending into the inner half of enamel, (3) caries in dentin and (4) deep dentin caries. Therefore, cut-off points can be defined as follow:

D1: all caries lesions are considered disease (1, 2, 3 and 4);

D2: only caries lesions from the inner half of enamel are considered disease (2, 3 and 4); D3,D4: only dentin caries lesions are considered disease (3 and 4).

0 0