However, a different pattern emerged from correlations of dichoptic ASD and dominant DSD ratios separately for each frequency segment (
Figs. 4E–H). These ratios showed significant correlations only for low, Spearman rho
(N = 29) = –0.57,
P = 0.0011, and lower middle spatial frequencies, rho
(N = 29) = –0.46,
P = 0.012, while there was no detectable correlation either in the upper middle, rho
(N = 29) = –0.17,
P = 0.39, or the high spatial frequency segments, rho
(N = 29) = –0.15,
P = 0.44. In contrast, group separation increased from lower to higher spatial frequencies, largely because of the marked decrease in amblyopic contrast sensitivity towards large spatial frequencies. K-means clustering analysis was quite accurate in dividing the group to controls and patients for the high (cluster 1,
n = 20; cluster 2,
n = 9; with three amblyopic subjects misplaced, two of whom have undergone binocular treatment) and upper middle spatial frequency segments (cluster 1,
n = 21 and cluster 2,
n = 8, with four amblyopic subjects misplaced, three of whom have undergone binocular treatment), whereas it demonstrated worse performance compared with the whole AULCSF analysis in the cases of the lower middle (seven patients and one control misplaced) and low spatial frequency segments (complete mix of controls and patients within clusters). Interestingly, investigation of the relationship between the high frequency dichoptic ASD and the lower-middle dominant eye DSD, those segments that have shown the biggest difference between groups, yielded a unique pattern (
Fig. 4G): the cluster analysis results were dominated by the high between-group separation of high frequency dichoptic ASD (cluster 1,
n = 20 and cluster 2,
n = 9; with three amblyopic subjects misplaced, two of whom have undergone binocular treatment), while the data points of the previously binocularly treated individuals lay between the two groups, bridging the gap between patients and controls, also creating a week correlative trend, which did not reach significance (rho
(N = 29) = –0.24,
P = 0.22). Taken together, these results suggest that the absence of DSD effect (i.e., dichoptic masking) in contrast sensitivity of the dominant eye in the low spatial frequency range could be an amblyopic feature to look for especially when evaluating the effectiveness of binocular treatment, other than the well-known and characteristic amblyopic contrast sensitivity deficit in the high spatial frequency range.