Skip to main content
. 2022 Aug 29;17(8):e0273029. doi: 10.1371/journal.pone.0273029

Table 4. Comparison of L, a* and b* values for images with post processing white balance correction using Macbeth color chart.

L
Median (IQR) χ2 (df) p* a
Spectrophotometer (control) 62.45 (7.80) 16.904 (4) p = 0.000
Windowless clinic 1 64.45 (9.60)
Windowless clinic 2 63.00 (7.38)
Windowed clinic 1 62.00 (11.15)
Windowed clinic 2 62.70 (13.25)
a*
Median (IQR) χ2 (df) P* b
Spectrophotometer (control) 7.00 (5.10) 8.933 (4) p = 0.015
Windowless clinic 1 6.75 (4.15)
Windowless clinic 2 6.60 (3.43)
Windowed clinic 1 6.80 (3.10)
Windowed clinic 2 6.80 (3.30)
b*
Median (IQR) χ2 (df) P* c
Spectrophotometer (control) 43.20 (10.68) 205.699 (4) p = 8.376 × 10−16
Windowless clinic 1 20.05 (3.58)
Windowless clinic 2 19.80 (3.20)
Windowed clinic 1 21.70 (3.38)
Windowed clinic 2 21.85 (3.25)

*Significant < 0.05; χ2 = Chi-square statistics; df = degree of freedom, IQR = Interquartile Range

*Kruskal-Wallis one-way test: Parametric assumption not met. Shapiro- Wilk test significant (p < 0.05)

a Post-hoc analysis (Dunn’s test) for L: Null hypothesis rejected (p = 0.002). Spectrophotometer Vs all other lighting conditions showed no significant differences (p > 0.05) except windowless clinic 1 (p < 0.001). Additionally, no significant differences observed for windowed clinic 1 vs windowed clinic 2 (p = 0.359), windowed clinic 1 vs windowless clinic 2 (p = 0.115), windowed clinic 2 vs windowless clinic 2 (p = 0.509), windowless clinic 1 vs windowless clinic 2 (p = 0.060)

b Post -hoc analysis (Dunn’s test) for a*: Null hypothesis accepted (p = 0.063)

c Post -hoc analysis (Dunn’s test) for b*: Null hypothesis rejected (p < 0.001). Spectrophotometer vs all other lighting conditions showed highly significant differences (p < 0.001). No significant differences observed between windowless clinic 1 vs windowless clinic 2 (p = 0.532) and windowed clinic 1 vs windowed clinic 2 (p = 0.953)