Claire Lifan Chen

Deep Learning in Label-free Cell Classification

Label-free cell analysis is essential to personalized genomics, cancer diagnostics, and drug development as it avoids adverse effects of staining reagents on cellular viability and cell signaling. However, currently available label-free cell assays mostly rely only on a single feature and lack sufficient differentiation. Also, the sample size analyzed by these assays is limited due to their low throughput. Here, we integrate feature extraction and deep learning with high-throughput quantitative imaging enabled by photonic time stretch, achieving record high accuracy in label-free cell classification. Our system captures quantitative optical phase and intensity images and extracts multiple biophysical features of individual cells. These biophysical measurements form a hyperdimensional feature space in which supervised learning is performed for cell classification. We compare various learning algorithms including artificial neural network, support vector machine, logistic regression, and a novel deep learning pipeline, which adopts global optimization of receiver operating characteristics. As a validation of the enhanced sensitivity and specificity of our system, we show classification of white blood T-cells against colon cancer cells, as well as lipid accumulating algal strains for biofuel production. This system opens up a new path to data-driven phenotypic diagnosis and better understanding of the heterogeneous gene expressions in cells.

Optical Data Compression in Time Stretch Imaging

Time stretch imaging offers real-time image acquisition at millions of frames per second and subnanosecond shutter speed, and has enabled detection of rare cancer cells in blood with record throughput and specificity. An unintended consequence of high throughput image acquisition is the massive amount of digital data generated by the instrument. Here we report the first experimental demonstration of real-time optical image compression applied to time stretch imaging. By exploiting the sparsity of the image, we reduce the number of samples and the amount of data generated by the time stretch camera in our proof-of-concept experiments by about three times. Optical data compression addresses the big data predicament in such systems.

Design of Warped Stretch Transform.

Time stretch dispersive Fourier transform enables real-time spectroscopy at the repetition rate of million scans per second. High-speed real-time instruments ranging from analog-to-digital converters to cameras and single-shot rare-phenomena capture equipment with record performance have been empowered by it. Its warped stretch variant, realized with nonlinear group delay dispersion, offers variable-rate spectral domain sampling, as well as the ability to engineer the time-bandwidth product of the signal’s envelope to match that of the data acquisition systems. To be able to reconstruct the signal with low loss, the spectrotemporal distribution of the signal spectrum needs to be sparse. Here, for the first time, we show how to design the kernel of the transform and specifically, the nonlinear group delay profile dictated by the signal sparsity. Such a kernel leads to smart stretching with nonuniform spectral resolution, having direct utility in improvement of data acquisition rate, real-time data compression, and enhancement of ultrafast data capture accuracy. We also discuss the application of warped stretch transform in spectrotemporal analysis of continuous-time signals.

Label-free high-throughput imaging flow cytometry

Flow cytometry is an optical method for studying cells based on their individual physical and chemical characteristics. It is widely used in clinical diagnosis, medical research, and biotechnology for analysis of blood cells and other cells in suspension. Conventional flow cytometers aim a laser beam at a stream of cells and measure the elastic scattering of light at forward and side angles. They also perform single-point measurements of fluorescent emissions from labeled cells. However, many reagents used in cell labeling reduce cellular viability or change the behavior of the target cells through the activation of undesired cellular processes or inhibition of normal cellular activity. Therefore, labeled cells are not completely representative of their unaltered form nor are they fully reliable for downstream studies. To remove the requirement of cell labeling in flow cytometry, while still meeting the classification sensitivity and specificity goals, measurement of additional biophysical parameters is essential. Here, we introduce an interferometric imaging flow cytometer based on the world’s fastest continuous-time camera. Our system simultaneously measures cellular size, scattering, and protein concentration as supplementary biophysical parameters for label-free cell classification. It exploits the wide bandwidth of ultrafast laser pulses to perform blur-free quantitative phase and intensity imaging at flow speeds as high as 10 meters per second and achieves nanometer-scale optical path length resolution for precise measurements of cellular protein concentration.

High-throughput biological cell classification featuring real-time optical data compression

High throughput real-time instruments are needed to acquire large data sets for detection and classification of rare events. Enabled by the photonic time stretch digitizer, a new class of instruments with record throughputs have led to the discovery of optical rogue waves [1], detection of rare cancer cells [2], and the highest analog-to-digital conversion performance ever achieved [3]. Featuring continuous operation at 100 million frames per second and shutter speed of less than a nanosecond, the time stretch camera is ideally suited for screening of blood and other biological samples. It has enabled detection of breast cancer cells in blood with record, one-in-a-million, sensitivity [2]. Owing to their high real-time throughput, instruments produce a torrent of data - equivalent to several 4K movies per second - that overwhelm data acquisition, storage, and processing operations. This predicament calls for technologies that compress images in optical domain and in real-time. An example of this, based on warped stretch transformation and non-uniform Fourier domain sampling will be reported.

Context-Aware Image Compression.

We describe a physics-based data compression method inspired by the photonic time stretch wherein information-rich portions of the data are dilated in a process that emulates the effect of group velocity dispersion on temporal signals. With this coding operation, the data can be downsampled at a lower rate than without it. In contrast to previous implementation of the warped stretch compression, here the decoding can be performed without the need of phase recovery. We present rate-distortion analysis and show improvement in PSNR compared to compression via uniform downsampling.

Time stretch imaging with optical data compression for label-free biological cell classification

Real-time instruments that acquire large data sets are needed for detection and classification of outliers. A new class of high throughput real-time instruments based on the photonic time-stretch has led to the discovery of optical rogue waves [1], detection of rare cancer cells [2], and the highest analog-to-digital conversion performance ever achieved [3]. One example of these instruments is the time stretch camera, an imaging modality that features continuous operation at about 100 million frames per second and shutter speed of less than a nanosecond. As an imaging flow-through microscope, the technology is in clinical testing for blood screening. While highly useful for collecting large data sets, the instruments ultrahigh throughput also creates a big data problem. The system produces a large volume of data in a short time equivalent to several 4K movies per second. Such a data fire hose places a burden on data acquisition, storage, and processing operations and calls for technologies that compress images in optical domain and in real-time. An example of this, based on warped stretch transformation and non-uniform Fourier domain sampling has recently been reported [4]. The paper will provide an overview of the time-stretch microscope with real-time optical image compression, and application of this technology in classification of cancer cell lines in blood.

AI-augmented time stretch microscopy

Cell reagents used in biomedical analysis often change behavior of the cells that they are attached to, inhibiting their native signaling. On the other hand, label-free cell analysis techniques have long been viewed as challenging either due to insufficient accuracy by limited features, or because of low throughput as a sacrifice of improved precision. We present a recently developed artificial-intelligence augmented microscope, which builds upon high-throughput time stretch quantitative phase imaging (TS-QPI) and deep learning to perform label-free cell classification with record high-accuracy. Our system captures quantitative optical phase and intensity images simultaneously by frequency multiplexing, extracts multiple biophysical features of the individual cells from these images fused, and feeds these features into a supervised machine learning model for classification. The enhanced performance of our system compared to other label-free assays is demonstrated by classification of white blood T-cells versus colon cancer cells and lipid accumulating algal strains for biofuel production, which is as much as five-fold reduction in inaccuracy. This system obtains the accuracy required in practical applications such as personalized drug development, while the cells remain intact and the throughput is not sacrificed. Here, we introduce a data acquisition scheme based on quadrature phase demodulation that enables interruptionless storage of TS-QPI cell images. Our proof of principle demonstration is capable of saving 40 TB of cell images in about four hours, i.e. pictures of every single cell in 10 mL of a sample.

Three-Dimensional Ultrafast Laser Scanner

Laser scanners are essential for scientific research, manufacturing, defense, and medical practice. Unfortunately, often times the speed of conventional laser scanners (e.g., galvanometric mirrors and acousto-optic deflectors) falls short for many applications, resulting in motion blur and failure to capture fast transient information. Here, we present a novel type of laser scanner that offers roughly three orders of magnitude higher scan rates than conventional methods. Our laser scanner, which we refer to as the hybrid dispersion laser scanner, performs inertia-free laser scanning by dispersing a train of broadband pulses both temporally and spatially. More specifically, each broadband pulse is temporally processed by time stretch dispersive Fourier transform and further dispersed into space by one or more diffractive elements such as prisms and gratings. As a proof-of-principle demonstration, we perform 1D line scans at a record high scan rate of 91 MHz and 2D raster scans and 3D volumetric scans at an unprecedented scan rate of 105 kHz. The method holds promise for a broad range of scientific, industrial, and biomedical applications. To show the utility of our method, we demonstrate imaging, nanometer-resolved surface vibrometry, and high-precision flow cytometry with real-time throughput that conventional laser scanners cannot offer due to their low scan rates.

Time Stretch Quantitative Phase Imaging

Label-free cell analysis is essential to personalized genomics, cancer diagnostics, and drug development as it avoids adverse effects of staining reagents on cellular viability and cell signaling. However, currently available label-free cell assays mostly rely only on a single feature and lack sufficient differentiation. Also, the sample size analyzed by these assays is limited due to their low throughput. Here, we integrate feature extraction and deep learning with high-throughput quantitative imaging enabled by photonic time stretch, achieving record high accuracy in label-free cell classification. Our system captures quantitative optical phase and intensity images and extracts multiple biophysical features of individual cells. These biophysical measurements form a hyperdimensional feature space in which supervised learning is performed for cell classification.