Review Article

Structural Study of Heterogeneous Biological Samples by Cryoelectron Microscopy and Image Processing

Figure 2

Multivariate Statistical Analysis. (a) Left: ten images, each consisting of 2 pixels. Right: each image is represented as a vector in 2-dimensional space according to their grey values. (b) Hierarchical Classification. The left panel shows the sequential combination of vectors according to their closeness. The initial classification of images starts by forming small classes which include images that are close to one another in multidimensional space and then the size of the group is progressively increased by merging with dimensional other surrounding smaller groups that are in close proximity to each other (see the text). Images that are too far from each other form new separate classes. In the example shown in panel (b) the process of forming two classes is represented by the blue and green ovals which have varying degrees of colour intensity. The light and dark coloured ovals correspond to the initial and final steps of classification, respectively. The right panel shows a tree of HAC. The starting point is 10 classes which correspond to the number of single images in the dataset. The cut-off point is shown by the dashed red line if 2 classes are required and this corresponds to the two classes shown in the left panel.
(a)
(b)