Pop out many small structures from a very large microscopic image

In medical research, many applications require counting and measuring small regions in a large image. Extracting these regions poses a dilemma in terms of segmentation granularity due to fine structures and segmentation complexity due to large image sizes. We propose a constrained spectral graph partitioning framework to address the former while also reducing the segmentation complexity associated with the latter. The final segmentation is obtained from a set of patch segmentations, each independently derived subject to stitching constraints between neighboring patches. Individual segmentation is based on local pairwise cues designed to pop out all cells simultaneously from their common background, while the constraints are derived from mutual agreement analysis on patch segmentations from a previous round of segmentation. Our results demonstrate that the constrained segmentation not only stitches solutions seamlessly along overlapping patch borders but also refines the segmentation in the patch interiors.

Algorithm overview for segmenting numerous small structures in a large image, performed as a series of independent patch segmentations subject to stitching constraints between them. The constraints are derived from mutual agreement analysis on adjacent patch segmentations from a previous round. The constrained segmentation not only stitches solutions seamlessly along overlapping patch borders but also refines the segmentation in the patch interiors.Figure optionsDownload high-quality image (221 K)Download as PowerPoint slideHighlights
► Simple spectral graph algorithm pops out many small regions all at once.
► Grouping cues based on short-range attraction and long-range repulsion between pixels.
► Efficient solution handles fine segmentation granularity and large image complexity.