The ratio ranged from 0 to 1.

Sample size was determined using the free software G* Power 3.1 [11]. We evaluated 30 patients, and the effect size was calculated from the mean, standard deviation, and correlation. Wilcoxon signed-rank sum test was chosen, and the significance level was set to 0.05. The result showed that a sample size of 26 to 75 patients would provide a power of at least 0.8 for the difference between the superior and inferior walls. For the difference between areas, 27 to 48 patients were needed to provide a power of at least 0.8. Thus, the sample size in our study, 86 patients, was considered sufficient.

Interobserver agreement was evaluated by weighted κ-statistics. A κ-value of 0–0.2 was considered poor agreement, 0.2–0.4 fair agreement, 0.4–0.6 moderate agreement, 0.6–0.8 substantial agreement, and 0.8–1.0 almost perfect agreement [12].

To compare the visibility ratio between the superior and inferior walls in each area, the Wilcoxon signed-rank test was used. Further, to compare the visibility ratio of each wall among the three areas, post hoc comparisons with Scheffe’s test to make multiple comparisons following Friedman’s test were used. Analysis was performed with statistical software, Ekuseru-Toukei 2008, v. 1.10 (Social Survey Research Information Co., Ltd., Tokyo, Japan). A p value of < 0.05 was considered statistically significant.