Spatio-temporal feature learning with reservoir computing for T-cell segmentation in live-cell $$\hbox {Ca}^{2+}$$ fluorescence microscopy

Abstract

Advances in high-resolution live-cell $$\hbox {Ca}^{2+}$$ Ca 2 + imaging enabled subcellular localization of early $$\hbox {Ca}^{2+}$$ Ca 2 + signaling events in T-cells and paved the way to investigate the interplay between receptors and potential target channels in $$\hbox {Ca}^{2+}$$ Ca 2 + release events. The huge amount of acquired data requires efficient, ideally automated image processing pipelines, with cell localization/segmentation as central tasks. Automated segmentation in live-cell cytosolic $$\hbox {Ca}^{2+}$$ Ca 2 + imaging data is, however, challenging due to temporal image intensity fluctuations, low signal-to-noise ratio, and photo-bleaching. Here, we propose a reservoir computing (RC) framework for efficient and temporally consistent segmentation. Experiments were conducted with Jurkat T-cells and anti-CD3 coated beads used for T-cell activation. We compared the RC performance with a standard U-Net and a convolutional long short-term memory (LSTM) model. The RC-based models (1) perform on par in terms of segmentation accuracy with the deep learning models for cell-only segmentation, but show improved temporal segmentation consistency compared to the U-Net; (2) outperform the U-Net for two-emission wavelengths image segmentation and differentiation of T-cells and beads; and (3) perform on par with the convolutional LSTM for single-emission wavelength T-cell/bead segmentation and differentiation. In turn, RC models contain only a fraction of the parameters of the baseline models and reduce the training time considerably.

DARTS: an open-source Python pipeline for Ca2+ microdomain analysis in live cell imaging data

“…DARTS will be continuously extended. Current developments comprise the autodetection of bead contacts, adapting state-of-the-art machine learning-based methods like in ( 25 ), as well as integration of further deconvolution and denoising approaches. Moreover, options for autodetection of rare shape normalization failures (see Figure 5 for examples) will also be incorporated.…”

A Survey on Reservoir Computing and its Interdisciplinary Applications Beyond Traditional Machine Learning

“…Biomedical ECG, EMG and MCG signal processing [209,210,211,212,213,214]. Medical images segmentation and classification [215].…”

“…In addition to DNA based RC, a medical image classification with distributed representations on cellular automata RC was reported by [257] (see cellular automata RC in Section 4). Besides, a recent study on spatio-temporal feature learning used RC model for T-cell consistent segmentation [215]. Instead of only applying a single reservoir, the model used multiple reservoirs for image segmentation and classification, where each reservoir focuses on a specific area of the image to obtain local interactions.…”

Predicting sea surface temperatures with coupled reservoir computers

“…Reservoir computers have been shown to be excellent at predicting complex dynamical systems (Ghosh et al, 2021;Pandey and Schumacher, 2020), regardless of the relative simplicity of the approach. They have even been shown to be proficient in the prediction of spatiotemporally complex systems, such as the Kuramoto-Sivashinsky partial differential equation (Vlachas et al, 2020) and cell segmentation (Hadaeghi et al, 2021). In our case, the use of coupled reservoir computers is introduced due to the large number of points on the map, making it computationally challenging to utilize a single large reservoir computer.…”