Automated tracking of label-free cells with enhanced recognition of whole tracks

Abstract

Migration and interactions of immune cells are routinely studied by time-lapse microscopy of in vitro migration and confrontation assays. To objectively quantify the dynamic behavior of cells, software tools for automated cell tracking can be applied. However, many existing tracking algorithms recognize only rather short fragments of a whole cell track and rely on cell staining to enhance cell segmentation. While our previously developed segmentation approach enables tracking of label-free cells, it still suffers from frequently recognizing only short track fragments. In this study, we identify sources of track fragmentation and provide solutions to obtain longer cell tracks. This is achieved by improving the detection of low-contrast cells and by optimizing the value of the gap size parameter, which defines the number of missing cell positions between track fragments that is accepted for still connecting them into one track. We find that the enhanced track recognition increases the average length of cell tracks up to 2.2-fold. Recognizing cell tracks as a whole will enable studying and quantifying more complex patterns of cell behavior, e.g. switches in migration mode or dependence of the phagocytosis efficiency on the number and type of preceding interactions. Such quantitative analyses will improve our understanding of how immune cells interact and function in health and disease.

Enhanced segmentation of label‐free cells for automated migration and interaction tracking

“…Moreover, our algorithm can be adjusted to any experimental data set with relative ease for two reasons: (i) AMITV‐v3 does not require data annotation, which is mandatory for learning‐based approaches [5] and (ii) AMIT‐v3 requires only two main parameters that can be interactively adjusted, that is, the thresholds for segmentation and for the Canny filter during tracking. Since AMIT‐v3 was developed for monitoring migration scenarios [8, 10] as well as host‐pathogen confrontation assays [9] of non‐mitotic cells, we did not expect it to achieve a top ranking in the CTC challenge on datasets where mother‐daughter association of dividing cells occurs, although it still performed satisfactorily. Furthermore, AMIT‐v3 proved to be the best method on the in‐house data compared to two established ML‐based approaches.…”

“…We advanced AMIT‐v1 [8, 9] to AMIT‐v2 [10] by improving the recognition of whole tracks. We briefly summarize the segmentation and tracking of cells by the following six steps: Regions of interest (ROIs) are detected and assigned by a GMM according to the spatiotemporal variances of pixel intensities to the classes: ‘background’, ‘static objects’, or ‘moving objects’. All ROIs of the class ‘moving objects’ are distinguished by a second GMM based on their area into one of the three classes: ‘noise’, ‘single cells’, or ‘cell clusters’. Cells can dynamically change their morphological state, for example, appearing as spreading cells [11, 12].…”

“…We advanced AMIT-v1 [8,9] to AMIT-v2 [10] by improving the recognition of whole tracks. We briefly summarize the segmentation and tracking of cells by the following six steps:…”

“…The first version of AMIT [8, 9]—referred to as AMIT‐v1—was extended by its second version—referred to as AMIT‐v2—for the enhanced recognition of cell tracks as a whole by the identification and improvement of various sources for track fragmentation [10]: (i) detection of spreading cells with low contrast and (ii) optimization of the gap size parameter value, which defines the acceptable number of missing cell positions between fragments of the same track. The enhanced track recognition in AMIT‐v2 yielded a significant increase in performance over AMIT‐v1; for example, the average length of cell tracks was increased by more than a factor 2 [10].…”

An example of in situ ductile damage analysis by tracking algorithm

“…The theory of the tracking algorithm is classical (also named Particle Image Velocimetry 10 ), and widely used in different fields: biology to follow cells evolution in situ using time-lapse videos, [11][12][13][14][15] structural timber engineering, 16 and much more... The algorithm used here is an extension of previous publications, 2,13 adapted to tomographic datasets.…”