FIG. 2 shows an exemplary biological specimen image 210 (in this example, an H&E image), and an exemplary plurality of superpixels 220 generated for image 210 by superpixel generator 110. Based on this example, it is appreciated that each superpixel can include the same number of pixels or a different number of pixels (e.g., within a certain range), but in either case the number of superpixels can be significantly (e.g., one or more orders of magnitude) lower than the number of pixels in image 210.

After superpixels have been generated by superpixel generator 110 (or otherwise obtained by system 100), the superpixels can be provided to feature extractor 111. Feature extractor 111 may extract from (or generate for) each superpixel a plurality of image features characterizing (or representing) the superpixel. As discussed above, the extracted image features may include, for example, texture features such Haralick features, bag-of-words features and the like. The values of the plurality of image features may be combined into a high-dimensional vector, hereinafter referred to as the “feature vector” characterizing the superpixel. For example, if M features are extracted for each superpixel, each superpixel can be characterized by an M-dimensional feature vector.