New method for internal anal sphincter measurements: feasibility study.

PURPOSE: The aim of this paper is to provide a method for measuring the internal anal sphincter on the basis of the quantitative analysis of three-dimensional endosonographic images. A software calculates a large set of measurements which are able to describe the three-dimensional shape of the muscle.
METHODS: A software provides four types of measurements: thickness, length, area and volume. The different magnitudes are estimated using the same reference system. The measurements obtained are modeled by functions that describe their spatial trend. The precision and reproducibility of the method was tested on a phantom before a study was performed on fifteen healthy patients. The measurements were carried out by two different operators. The inter-observer variability were assessed.
RESULTS: In the phantom measurements the mean errors and the standard deviation were: 0.05 +/- 0.1 mm for the thickness, 0.02 +/- 0.12 mm for the length, -4.43 +/- 2.4 mm(2) for the area, -20.69 +/- 20.83 mm(3) for the volume. The maximum absolute differences between the measurements carried out by the two operators was: 0.18 mm for the thickness (in the 95% of the case), 1.69 mm(2) for the area (in the 95% of the case), and 0.25 mm for the length, and 29.46 mm(3) for the volume. The human IAS assessments were evaluated on each segment. The mean of the all tissue measurements carried out were (mean +/- SD): 1.71 +/- 0.34 mm for the thickness, 33.24 +/- 6.10 mm for the length, 111.28 +/- 29.08 mm(2) for the area. The mean of the volume measurements of the entire tissue was: 4124 +/- 1160 mm(3). Inter-observer variability was observed only in the anterior proximal segment for the thickness measurements by Wilcoxon's signed rank test (P value = 0.048) and for the volume assessments by the limits of agreement method (-118 to 78 mm(3)). The mean percentage errors and the limit of agreement for the measurements of the entire tissue were: 0.27 and (-0.11 to 0.12 mm) for the thickness, -2.32 and (-3.88 to 2.33 mm) for the length, -0.05 and (-9.71 to 9.83 mm(2)) for the area, -1.89 and (-366 to 240 mm(3)) for the volume.
CONCLUSION: The assessments of accuracy and precision of the method result satisfactory for all four type of measurements. The reproducibility analysis confirms very good inter-observer agreement for the phantom measurements and for the most part of the IAS segments evaluations. Inter-observer variability was seen only for the thickness and volume measurements of the anterior-proximal segment. Our method provides a high number of measurements with good accuracy enabling a very detailed study of IAS morphology.