Comparison of reproducibility of different contour methods in measuring vascularization of cervical carcinoma with VOCAL system

Abstract

Objective To assess the reproducibility of vascularization measurement in cervical carcinoma using transvaginal three-dimensional power Doppler angiography (3D-PDA) with virtualorgan computer-aided analysis ( VOCAL), and compare the reproducibility of two different contour mode (manual and automatic mode) of VOCAL. Methods Eighty patients with cervical carcinoma were examined by observer 1 using transvaginal 3D-PDA. The two acquired volume datasets were analyzed using manual and sphere automatic contour mode of the VOCAL imaging program for assessing carcinoma vascularization index( VI), flow index(FI), and vascularization flow index(VFI). Forty patients of them were examined randomly by observer 2 by the same method. Reproducibility of vascularity measurement was assessed by calculating intraclass (intra-CC) and interclass (inter-CC) , limits of agreement, 95% confidence interval definition. The contribution of various factors (examiner, measurement, contour mode and patient) to intrasubject variance was estimated using different analysis of variance models (ANOVA). Results For intraobserver,manual contour mode was more valid than sphere automatic contour mode for each observer (0. 92～0. 99 vs 0. 75～0. 94) ,and it also had smaller standard deviation,narrower limit of agreement range, 95% confidence interval and higher intra-cc than those of sphere automatic contour mode. Interobserver agreement of manual contour vascular measurements was similar to the intraobserver agreement for manual contour vascular measurements ( 0. 89 ～ 0. 97 vs 0. 92 ～ 0. 99 ) , but interobserver agreement for sphere automatic contour vascular measurements dramatically reduced (0. 52～0. 72). Conclusions Manual contour mode for 3D-PDA vascular measurement has better reproduciblity than sphere automatic contour mode, especially useful for irregular shape tissue. Key words: Ultrasonographys Uterine cervical neoplasms; Virtual organ computer-aided analysis; Reproducibility of results