Dataset Information

Deep Cytometry: Deep learning with Real-time Inference in Cell Sorting and Flow Cytometry.

ABSTRACT: Deep learning has achieved spectacular performance in image and speech recognition and synthesis. It outperforms other machine learning algorithms in problems where large amounts of data are available. In the area of measurement technology, instruments based on the photonic time stretch have established record real-time measurement throughput in spectroscopy, optical coherence tomography, and imaging flow cytometry. These extreme-throughput instruments generate approximately 1 Tbit/s of continuous measurement data and have led to the discovery of rare phenomena in nonlinear and complex systems as well as new types of biomedical instruments. Owing to the abundance of data they generate, time-stretch instruments are a natural fit to deep learning classification. Previously we had shown that high-throughput label-free cell classification with high accuracy can be achieved through a combination of time-stretch microscopy, image processing and feature extraction, followed by deep learning for finding cancer cells in the blood. Such a technology holds promise for early detection of primary cancer or metastasis. Here we describe a new deep learning pipeline, which entirely avoids the slow and computationally costly signal processing and feature extraction steps by a convolutional neural network that directly operates on the measured signals. The improvement in computational efficiency enables low-latency inference and makes this pipeline suitable for cell sorting via deep learning. Our neural network takes less than a few milliseconds to classify the cells, fast enough to provide a decision to a cell sorter for real-time separation of individual target cells. We demonstrate the applicability of our new method in the classification of OT-II white blood cells and SW-480 epithelial cancer cells with more than 95% accuracy in a label-free fashion.

SUBMITTER: Li Y

PROVIDER: S-EPMC6668572 | biostudies-literature | 2019 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep Cytometry: Deep learning with Real-time Inference in Cell Sorting and Flow Cytometry.

Li Yueqin Y Mahjoubfar Ata A Chen Claire Lifan CL Niazi Kayvan Reza KR Pei Li L Jalali Bahram B

Scientific reports 20190731 1

Deep learning has achieved spectacular performance in image and speech recognition and synthesis. It outperforms other machine learning algorithms in problems where large amounts of data are available. In the area of measurement technology, instruments based on the photonic time stretch have established record real-time measurement throughput in spectroscopy, optical coherence tomography, and imaging flow cytometry. These extreme-throughput instruments generate approximately 1 Tbit/s of continuo ...[more]

PMID: 31366998

Similar Datasets

Project description:Electronic cell sorting for isolation and culture of dinoflagellates and other marine eukaryotic phytoplankton was compared to the traditional method of manually picking cells using a micropipette. Trauma to electronically sorted cells was not a limiting factor, as fragile dinoflagellates, such as Karenia brevis (Dinophyceae), survived electronic cell sorting to yield viable cells. The rate of successful isolation of large-scale (> 4 litres) cultures was higher for manual picking than for electronic cell sorting (2% vs 0.5%, respectively). However, manual picking of cells is more labor intensive and time consuming. Most manually isolated cells required repicking, as the cultures were determined not to be unialgal after a single round of isolation; whereas, no cultures obtained in this study from electronic single-cell sorting required resorting. A broad flow cytometric gating logic was employed to enhance species diversity. The percentages of unique genotypes produced by manual picking or electronic cell sorting were similar (57% vs 54%, respectively), and each approach produced a variety of dinoflagellate or raphidophyte genera. Alternatively, a highly restrictive gating logic was successfully used to target K. brevis from a natural bloom sample. Direct electronic single-cell sorting was more successful than utilizing a pre-enrichment sort followed by electronic single-cell sorting. The appropriate recovery medium may enhance the rate of successful isolations. Seventy percent of isolated cells were recovered in a new medium (RE) reported here, which was optimized for axenic dinoflagellate cultures. The greatest limiting factor to the throughput of electronic cell sorting is the need for manual postsort culture maintenance and assessment of the large number of isolated cells. However, when combined with newly developed automated methods for growth screening, electronic single-cell sorting has the potential to accelerate the discovery of new algal strains.

Dataset Information

Deep Cytometry: Deep learning with Real-time Inference in Cell Sorting and Flow Cytometry.

Publications

Deep Cytometry: Deep learning with Real-time Inference in Cell Sorting and Flow Cytometry.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets