The statistical effects of measuring myocytes with different image zoom rates

Chengxin Li, Maiying Kong, Dongfeng Wu
Department of Bioinformatics and Biostatistics, School of Public Health and Informatics Science, University of Louisville, Louisville, KY, USA

Abstract: Quantitative image measurements are widely used in both basic and clinical medical research. The measurements can be taken under different image zoom rates. Usually, increasing the image zoom rate increases the accuracy of the measurement; however, increasing the image zoom rate also increases the measurement cost. We measured 218 rat myocytes (112 myocytes from treated rat heart; 106 from placebo rat heart) for image measurements of cross-sectional areas, under different image zoom rates, 75%, 100%, 150%, 200%, and 300%. The linear mixed-effects model, Student's t-test, and F test were used to investigate whether image measurements significantly change over different zoom rates, and to find the optimal zoom rate in terms of maximizing the test statistic while reducing measurement cost. We found that the means of the measured myocyte areas decreased with increased zoom rates for both the treated heart and the placebo heart. The means of measured areas at zoom rate 75% versus 100% were not significantly different for both the treated heart and the placebo heart. However, there was a significant decrease when the zoom rate increased from 100% to 150% for the treated heart and the placebo heart, and the amount of decrease of the means for the two types of hearts was different. When the zoom rate was increased from 150% to 200% and 300%, the mean of the measurements decreased. However, the amount of decrease for both the treated heart and the placebo heart was the same. When image zoom rate increases, measurement precision remains the same. The t-test statistics for comparing treated heart and placebo heart remain the same when zoom rate reaches 150% and above. Therefore, the image zoom rate at 150% is the optimal choice in terms of maximizing the test statistic and reducing the measurement cost.

Keywords: image zoom rate, myocyte, cross-sectional area measurement, measurement cost, wheat germ agglutinin (WGA) staining, linear mixed-effects model