Fig. 3: Importance of vascular features, classification and subgroup analysis for the microangiopathy score.

a, ROC curve of the developed classification/diabetes detection scheme with the selected threshold (0.65) of the microangiopathy score. The blue area indicates the 95% CI of ROC curves for multiple iterations. b, Relative importance of the 32 selected 2D and 3D RSOM-extracted vascular features for the calculation of the microangiopathy score. c, Classification of healthy volunteers and patients with diabetes based on the calculated microangiopathy score. d, Efficacy metrics of the proposed method for diabetes detection. e, Statistical description of the groups of macroscale, mesoscale and microscale vascular features based on their relative importance for diabetes detection by means of the microangiopathy score. f, Comparison of the mean feature segmental maps among the healthy and diabetes groups in all scales of detail. The intensity of each segment corresponds to the normalized mean value of the feature for each group, as described by the three colour bars (blue, orange, grey). g, Statistical comparison of the normalized values, against the maximum value on a per-feature basis, of the selected features between healthy volunteers and patients for all three scales of detail. All P values were calculated using a two-sided Student’s t-test. For the boxplots, the centre line represents the median, the box limits the first and third quartiles, and the whiskers (minima and maxima) 1.5× the interquartile ranges.