A comparative analysis of manual point-counting method and automated pixel-counting method of area percentage estimation on placenta.

摘要

OBJECTIVE: To obtain the coefficient of variation (CV) of systematic point counting performed on sections of placenta; to obtain the optimal minimum CV achievable by systematic point-counting method for placentas; and to compare the manual point-counting and automatic pixel-counting methods and obtain a mean bias and 95% limits of agreement between them. STUDY DESIGN: This study used sections from 10 placentas. Two histopathologists used the point-counting method to calculate the area percentage occupied by chorionic villi by overlaying 25, 50, 100, 200, 300, 400, 500 and 1,000 points. Area percentage was also calculated by the automated pixel-counting method. RESULTS: The interobserver difference in the point-counting method decreased as the number of points overlaid increased and almost plateaued after an overlay of 300 points. CV was 2.2% and 1.1% when 300 and 1,000 points were overlaid, respectively. The deviation of results obtained by both methods reduced as the number of points overlaid increased. Additionally, there was a strong correlation between both methods when 300 points were overlaid. CONCLUSION: Interobserver concordance in the point-counting technique is reasonably high and the results obtained by the point-counting method approach those seen by the automated pixel-counting when at least 300 points are overlaid.