Reproducibility of two different methods for performing mean gray value evaluation of cyst content in endometriomas using VOCAL
Stefano Guerriero 1 , Juan Luis Alcazar 2 , Monica Pilloni 3 , Silvia Ajossa 3 , Begoña Olartecoechea 2 , Federica Sedda 3 , Alba Piras 3 , Gian Benedetto Melis 3 , Luca Saba 4
Objective: To compare two different methods (manual sampling of the entire cyst and semi-automated spherical sampling from the central part of the cyst) for calculating the mean gray value (MGV) from the cystic content in endometriomas using virtual organ computer-aided analysis (VOCAL).
Methods: Forty-one volumes from histologically confirmed endometriomas were retrieved from our database and the volumes were analyzed to compare the MGVs obtained via the two modalities. In addition, to evaluate the reproducibility in a sample of 20 volumes, two different observers calculated the MGV from cyst content using VOCAL software. For each method, each examiner analyzed the volumes twice, 3 weeks apart, for assessment of intra-observer agreement.
First, manual sampling of the internal contour of all the cysts was performed, and 1 week later semi-automated 2-cm sphere sampling from the central part of the cyst was carried out. In addition, the observers recorded the time spent performing each analysis. Inter- and intra-observer reproducibility was evaluated for each method using intra-class correlation coefficients (ICC).
Results: There was no difference in the mean MGV between manual sampling (22.211 ± 7.541) and the semi-automated modality of sampling (23.840 ± 8.621, p = 0.439). The correlation between manual and semi-automated sampling measurement was high (r = 0.92). According to the ICCs, there was no significant difference in interobserver reliability between manual sampling (0.931; 95 % CI, 0.824-0.973) and the semi-automated modality of sampling (0.924; 95 % CI, 0.809-0.970).
Intra-observer reproducibility for both examiners was good (ICC > 0.94). Semi-automated measurements were obtained faster than those obtained by manual evaluation (p = 0.0001 for observer 1 and p = 0.083 for observer 2).
Conclusions: Both methods seem to be reliable, but the semi-automated method using the sphere should be preferred because it is a less time-consuming procedure.