The accuracy of a postprocessing technique--volume render mode--in three-dimensional endoanal ultrasonography of anal abscesses and fistulas.

OBJECTIVE

This study aimed to evaluate the accuracy of a 3-dimensional volume render mode endoanal ultrasonography in the assessment of anal abscesses and fistulas.

PATIENTS AND METHODS

Three-dimensional endoanal ultrasonography was performed preoperatively in 62 patients with clinical diagnoses of an anal abscess and/or an anal fistula. The accuracy of a volume render mode endoanal ultrasonography was evaluated blindly by 2 observers through a retrospective analysis of all the stored images of anal abscesses and fistulas. Differences in classification of the type of anal abscesses and in determination of the type and location of internal openings of anal fistulas were assessed.

RESULTS

In classification of the primary tract, the agreement between the volume render mode endoanal ultrasonography and surgery was much higher than that for the 3-dimensional endoanal ultrasonography (marginal homogeneity test P > .1, κ = 0.96, and marginal homogeneity test P = .0048, κ = 0.28, respectively). In localizing the internal openings and in classifying anal abscesses, the agreement for both methods with surgery was identical. The interobserver agreement showed complete agreement (100%) with regard to luminance and thickness parameters and very high accuracy for the opacity and filter parameters.

CONCLUSIONS

This study demonstrated that volume render mode endoanal ultrasonography is a useful adjunct method to the 3-dimensional endoanal ultrasonography, showing better classification results of the types of anal fistulas.