Search

An unresolved issue in contemporary biomedicine is the overwhelming number and diversity of complex images that require annotation, analysis and interpretation. Recent advances in Deep Learning have revolutionized the field of computer vision, creating algorithms that compete with human experts in image segmentation tasks. However, these frameworks require large human-annotated datasets for training and the resulting "black box" models are difficult to interpret. In this study, we introduce Kartezio, a modular Cartesian Genetic Programming-based computational strategy that generates fully transparent and easily interpretable image processing pipelines by iteratively assembling and parameterizing computer vision functions. The pipelines thus generated exhibit comparable precision to state-of-the-art Deep Learning approaches on instance segmentation tasks, while requiring drastically smaller training datasets. This Few-Shot Learning method confers tremendous flexibility, speed, and functionality to this approach. We then deploy Kartezio to solve a series of semantic and instance segmentation problems, and demonstrate its utility across diverse images ranging from multiplexed tissue histopathology images to high resolution microscopy images. While the flexibility, robustness and practical utility of Kartezio make this fully explicable evolutionary designer a potential game-changer in the field of biomedical image processing, Kartezio remains complementary and potentially auxiliary to mainstream Deep Learning approaches. Deep learning frameworks require large human-annotated datasets for training and the resulting 'black box' models are difficult to interpret. Here, the authors present Kartezio; a modular Cartesian Genetic Programming-based computational strategy that generates fully transparent and easily interpretable image processing pipelines. [ABSTRACT FROM AUTHOR]

Background: Cancer cell aggregation is a key process involved in the formation of tumor cell clusters. It has recently been shown that clusters of circulating tumor cells (CTCs) have an increased metastatic potential compared to isolated circulating tumor cells. Several widely used chemotherapeutic agents that target the cytoskeleton microtubules and cause cell cycle arrest at mitosis have been reported to modulate CTC number or the size of CTC clusters. Results: In this study, we investigated in vitro the impact of mitotic arrest on the ability of breast tumor cells to form clusters. By using live imaging and quantitative image analysis, we found that MCF-7 cancer cell aggregation is compromised upon incubation with paclitaxel or vinorelbine, two chemotherapeutic drugs that target microtubules. In line with these results, we observed that MCF-7 breast cancer cells experimentally synchronized and blocked in metaphase aggregated poorly and formed loose clusters. To monitor clustering at the single-cell scale, we next developed and validated an in vitro assay based on live video-microscopy and custom-designed micro-devices. The study of cluster formation from MCF-7 cells that express the fluorescent marker LifeAct-mCherry using this new assay allowed showing that substrate anchorage-independent clustering of MCF-7 cells was associated with the formation of actin-dependent highly dynamic cell protrusions. Metaphase-synchronized and blocked cells did not display such protrusions, and formed very loose clusters that failed to compact. Conclusions: Altogether, our results suggest that mitotic arrest induced by microtubule-targeting anticancer drugs prevents cancer cell clustering and therefore, could reduce the metastatic potential of circulating tumor cells. [ABSTRACT FROM AUTHOR]