Separating Touching Cells Using Pixel Replicated Elliptical Shape Models

Abstract

One of the most important and error-prone tasks in biological image analysis is the segmentation of touching or overlapping cells. Particularly for optical microscopy, including transmitted light and confocal fluorescence microscopy, there is often no consistent discriminative information to separate cells that touch or overlap. It is desired to partition touching foreground pixels into cells using the binary threshold image information only, and optionally incorporating gradient information. The most common approaches for segmenting touching and overlapping cells in these scenarios are based on the watershed transform. We describe a new approach called pixel replication for the task of segmenting elliptical objects that touch or overlap. Pixel replication uses the image Euclidean distance transform in combination with Gaussian mixture models to better exploit practically effective optimization for delineating objects with elliptical decision boundaries. Pixel replication improves significantly on commonly used methods based on watershed transforms, or based on fitting Gaussian mixtures directly to the thresholded image data. Pixel replication works equivalently on both 2-D and 3-D image data, and naturally combines information from multichannel images. The accuracy of the proposed technique is measured using both the segmentation accuracy on simulated ellipse data and the tracking accuracy on validated stem cell tracking results extracted from hundreds of live-cell microscopy image sequences. Pixel replication is shown to be significantly more accurate compared to other approaches. Variance relationships are derived, allowing a more practically effective Gaussian mixture model to extract cell boundaries for data generated from the threshold image using the uniform elliptical distribution and from the distance transform image using the triangular elliptical distribution.

Spatio-temporal Control of ERK Pulse Frequency Coordinates Fate Decisions during Mammary Acinar Morphogenesis

“…After denoising, a new ensemble-based segmentation algorithm was applied. This ensemble segmentation combined an adaptive thresholding into foreground/background regions with an anisotropic 3-D Laplacian of Gaussian filter targeted to a specific cell radius to separate touching cells (Winter et al, 2019). The base segmentation was run at different cell radii and the results were combined using unsupervised learning techniques from the field of algorithmic information theory (Cohen et al, 2009).…”

Artificial Intelligence-Powered Automated Holotomographic Microscopy Enables Label-Free Quantitative Biology

“…However, due to the heterogeneous nature of the material they stain (the cell's dry mass), their signal may show spatial and temporal variations making simple thresholding a challenge. In addition, the complex curvature of cell edges makes it difficult to distinguish the boundaries of touching cells, resulting in under-segmentation problems such as two or more cells being recognized as one, which presents a more complex problem, similar to the challenge posed by touching nuclei [22].…”

The cluster structure function

“…In the computational biological microscopy image analysis area we build on previous work for optimally partitioning connected components of foreground pixels into elliptical regions [34].…”

“…HSC colonies, or groups of touching cells, consist of dividing and differentiating cells that present a wide variety of sizes and shapes. The large morphological variation arises from both the presence of cells in developmental states and the mechanical interaction among adjacent cells deforming their shape, texture, and behavior [31], [32]. Timelapse microscopy of living cells further complicates the problem, requiring reduced imaging energy to lessen phototoxicity, and also introducing temporal variations due to imaging as well as cell and colony appearance variability.…”