Classification of Visual Cortex Plasticity Phenotypes following Treatment for Amblyopia

Figure 7

Clustering of samples with similar plasticity features identified using -distributed stochastic neighbor embedding (tSNE) and -means clustering. (a) Using tSNE to visualize clustering of samples (109 tissue samples from animals reared to 5 wk normal, 5 wk MD, RO, BD, and BV) calculated from -means analysis of the 8 plasticity features identified by PCA. The optimal number of clusters () was identified by measuring the within groups sum of squares at intervals between 2 and 9 clusters (Figure 7-1). (b) The content of each cluster was visualized for the region (central, peripheral, and monocular) (c) or treatment condition. (d) The table summarizes the percentage of samples for each region and condition in clusters 1-6. For example, 100% of the samples from the central region of the V1 in normal animals were in cluster 1 while 100% of the samples from all regions of RO were in cluster 2. This information was used to annotate subclusters based on the cluster membership (1-6), rearing condition, and region of the V1.

(a)

(b)

(c)

(d)