Table 2 Comparison of Average Thresholds by Each Thresholding Method, Stratified by Treatment.

From: A Sensitive Thresholding Method for Confocal Laser Scanning Microscope Image Stacks of Microbial Biofilms

Pairwise Thresholding Method Comparison Control Images (n = 25) Treatment Images (n = 25)
Mean Threshold Group 1 - Group 2 (95% confidence interval) Paired t-test p-value Mean Threshold Group 1 - Group 2 (95% confidence interval) Paired t-test p-value
Manual avga. vs BEM 28.69/11.36 17.33 (14.47, 20.19) <0.01 30.02/11.96 18.06 (16.09, 20.02) <0.01
Manual avg. vs Otsu 28.69/64.52 −35.82 (−39.44, −32.22) <0.01 30.02/65.28 −35.26 (−38.48, −32.04) <0.01
Manual avg. vs IS 28.69/66.40 −37.71 (−41.30, −34.13) <0.01 30.02/67.28 −38.26 (−40.45, −34.08) <0.01
BEM vs Otsu 11.36/64.52 −53.16 (−59.01, −47.31) <0.01 11.96/65.28 −53.32 (−57.89, −48.75) <0.01
BEM vs IS 11.36/66.40 −55.04 (−60.87, −49.21) <0.01 11.96/67.28 −55.32 (−59.85, −50.79) <0.01
Otsu vs IS 64.52/66.40 −1.88 (−2.02, −1.74) <0.01 65.28/67.28 −2.00 (−2.12, −1.88) <0.01
  1. Fifty oral biofilms were thresholded with four different methods. Half the biofilm images had been treated with water 8 and 18 hours into their 22 hour development and were designated as treatment biofilms. The other half were developed undisturbed over 22 hours and designated as control biofilms. The null hypothesis states that the mean difference between sets of thresholds obtained from one method vs another method is zero. Since all 12 null hypotheses were rejected, we conclude that each thresholding method was different from one another and is unaffected by treatment status of the images operated on. Although mean thresholds for Otsu and IS were roughly 2 intensity values apart, IS thresholds were consistently 2 units higher than Otsu thresholds applied to the same image, minimizing standard deviation and producing significant effects.
  2. aFor an image’s individual manual threshold value, the five values given by our five operators were averaged.