Benchmark for multi-cellular segmentation of bright field microscopy images

Bio-EdIP: An automatic approach for in vitro cell confluence images quantification

BACKGROUND AND OBJECTIVES

Cell imaging is a widely-employed technique to analyze multiple biological processes. Therefore, simple, accurate and quantitative tools are needed to understand cellular events. For this purpose, Bio-EdIP was developed as a user-friendly tool to quantify confluence levels using cell culture images.

METHODS

The proposed algorithm combines a pre-processing step with subsequent stages that involve local processing techniques and a morphological reconstruction-based segmentation algorithm. Segmentation performance was assessed in three constructed image sets, comparing F-measure scores and AUC values (ROC analysis) for Bio-EdIP, its previous version and TScratch. Furthermore, segmentation results were compared with published algorithms using eight public benchmarks.

RESULTS

Bio-EdIP automatically segmented cell-free regions from images of in vitro cell culture. Based on mean F-measure scores and ROC analysis, Bio-EdIP conserved a high performance regardless of image characteristics of the constructed dataset, when compared with its previous version and TScratch. Although acquisition quality of the public dataset affected Bio-EdIP segmentation, performance was better in two out of eight public sets.

CONCLUSIONS

Bio-EdIP is a user-friendly interface, which is useful for the automatic analysis of confluence levels and cell growth processes using in vitro cell culture images. Here, we also presented new manually annotated data for algorithms evaluation.

Super-resolved 3-D imaging of live cells' organelles from bright-field photon transmission micrographs

Current biological and medical research is aimed at obtaining a detailed spatiotemporal map of a live cell's interior to describe and predict cell's physiological state. We present here an algorithm for complete 3-D modelling of cellular structures from a z-stack of images obtained using label-free wide-field bright-field light-transmitted microscopy. The method visualizes 3-D objects with a volume equivalent to the area of a camera pixel multiplied by the z-height. The computation is based on finding pixels of unchanged intensities between two consecutive images of an object spread function. These pixels represent strongly light-diffracting, light-absorbing, or light-emitting objects. To accomplish this, variables derived from Rényi entropy are used to suppress camera noise. Using this algorithm, the detection limit of objects is only limited by the technical specifications of the microscope setup-we achieve the detection of objects of the size of one camera pixel. This method allows us to obtain 3-D reconstructions of cells from bright-field microscopy images that are comparable in quality to those from electron microscopy images.

Segmentation of Image Data from Complex Organotypic 3D Models of Cancer Tissues with Markov Random Fields

Organotypic, three dimensional (3D) cell culture models of epithelial tumour types such as prostate cancer recapitulate key aspects of the architecture and histology of solid cancers. Morphometric analysis of multicellular 3D organoids is particularly important when additional components such as the extracellular matrix and tumour microenvironment are included in the model. The complexity of such models has so far limited their successful implementation. There is a great need for automatic, accurate and robust image segmentation tools to facilitate the analysis of such biologically relevant 3D cell culture models. We present a segmentation method based on Markov random fields (MRFs) and illustrate our method using 3D stack image data from an organotypic 3D model of prostate cancer cells co-cultured with cancer-associated fibroblasts (CAFs). The 3D segmentation output suggests that these cell types are in physical contact with each other within the model, which has important implications for tumour biology. Segmentation performance is quantified using ground truth labels and we show how each step of our method increases segmentation accuracy. We provide the ground truth labels along with the image data and code. Using independent image data we show that our segmentation method is also more generally applicable to other types of cellular microscopy and not only limited to fluorescence microscopy.

How much information can be obtained from tracking the position of the leading edge in a scratch assay?

Moving cell fronts are an essential feature of wound healing, development and disease. The rate at which a cell front moves is driven, in part, by the cell motility, quantified in terms of the cell diffusivity D, and the cell proliferation rate λ. Scratch assays are a commonly reported procedure used to investigate the motion of cell fronts where an initial cell monolayer is scratched, and the motion of the front is monitored over a short period of time, often less than 24 h. The simplest way of quantifying a scratch assay is to monitor the progression of the leading edge. Use of leading edge data is very convenient because, unlike other methods, it is non-destructive and does not require labelling, tracking or counting individual cells among the population. In this work, we study short-time leading edge data in a scratch assay using a discrete mathematical model and automated image analysis with the aim of investigating whether such data allow us to reliably identify D and λ. Using a naive calibration approach where we simply scan the relevant region of the (D, λ) parameter space, we show that there are many choices of D and λ for which our model produces indistinguishable short-time leading edge data. Therefore, without due care, it is impossible to estimate D and λ from this kind of data. To address this, we present a modified approach accounting for the fact that cell motility occurs over a much shorter time scale than proliferation. Using this information, we divide the duration of the experiment into two periods, and we estimate D using data from the first period, whereas we estimate λ using data from the second period. We confirm the accuracy of our approach using in silico data and a new set of in vitro data, which shows that our method recovers estimates of D and λ that are consistent with previously reported values except that that our approach is fast, inexpensive, non-destructive and avoids the need for cell labelling and cell counting.

Are in vitro estimates of cell diffusivity and cell proliferation rate sensitive to assay geometry?

Cells respond to various biochemical and physical cues during wound-healing and tumour progression. in vitro assays used to study these processes are typically conducted in one particular geometry and it is unclear how the assay geometry affects the capacity of cell populations to spread, or whether the relevant mechanisms, such as cell motility and cell proliferation, are somehow sensitive to the geometry of the assay. In this work we use a circular barrier assay to characterise the spreading of cell populations in two different geometries. Assay 1 describes a tumour-like geometry where a cell population spreads outwards into an open space. Assay 2 describes a wound-like geometry where a cell population spreads inwards to close a void. We use a combination of discrete and continuum mathematical models and automated image processing methods to obtain independent estimates of the effective cell diffusivity, D, and the effective cell proliferation rate, λ. Using our parameterised mathematical model we confirm that our estimates of D and λ accurately predict the time-evolution of the location of the leading edge and the cell density profiles for both assay 1 and assay 2. Our work suggests that the effective cell diffusivity is up to 50% lower for assay 2 compared to assay 1, whereas the effective cell proliferation rate is up to 30% lower for assay 2 compared to assay 1.