Fig. 5

NeuriteNet effectively classifies images of DRGNs with difference in longest neurite. A, B Representative images of rDRGNs treated with Nocodazole Before (A) and After replating (B). C, D Same images as in A, B that were correctly classified as Before C or After D. The intensity of the color indicates the relative importance of that area. Blue and orange indicate areas that were used by NeuriteNet to suggest the image belonged to Before and After groups, respectively. E, F The representative images of rDRGNs (A, B) with their saliency map overlayed (C, D. G Comparison of performance of different classification approaches. The percentage of total images or traces (n = 178, 178, 255) classified correctly as belonging to Before or After treatment groups is shown along with kappa statistic. H Fractional distribution of Predicted Treatment Score. Color represents the actual group (after or before) to which the image corresponds. Of note, NeuriteNet classifies an image as “After” if the Predicted Treatment Score is less than 0.5 and to “Before” if more than 0.5. NeuriteNet classified most images correctly (the small orange bar at Predicted Treatment Score of 0.85 (appears brown as it is overlaying the blue) represents a small fraction of After images falsely classified as Before). I Scatter plot of rDRGN plotted by their length score and length using data normalized with a quantile transformation. Linear regression shows a modest correlation with the Predicted Treatment Score and length measurement (ρ = 0.39). Scale bar = 100 µm