Time Stretch Microscopy also known as Serial time-encoded amplified imaging/microscopy or stretched time-encoded amplified imaging/microscopy' (STEAM) is a fast real-time optical imaging method that provides MHz frame rate, ~100 ps shutter speed, and ~30 dB (× 1000) optical image gain. Based on the Photonic Time Stretch technique, STEAM holds world records for shutter speed and frame rate in continuous real-time imaging. STEAM employs the Photonic Time Stretch with internal Raman amplification to realize optical image amplification to circumvent the fundamental trade-off between sensitivity and speed that affects virtually all optical imaging and sensing systems. This method uses a single-pixel photodetector, eliminating the need for the detector array and readout time limitations. Avoiding this problem and featuring the optical image amplification for dramatic improvement in sensitivity at high image acquisition rates, STEAM's shutter speed is at least 1000 times faster than the state - of - the - art CCD and CMOS cameras. Its frame rate is 1000 times faster than fastest CCD cameras and 10 - 100 times faster than fastest CMOS cameras.
History
Time stretch microscopy and its application to microfluidics for classification of biological cells were invented at UCLA. It combines concept of spectrally encoded illumination with the photonic time stretch, an ultrafast realtime data acquisition technology developed earlier in the same lab to create a femtosecond real-time single-shot digitizer, and a single shot stimulated Raman spectrometer. The first demonstration was a one-dimensional version and later a two-dimensional version. Later, a fast imaging vibrometer was created by extending the system to an interferometric configuration. The technology was then extended to time stretch quantitive phase imaging (TS-QPI) for label free classification of blood cells and combined with artificial intelligence (AI) for classification of cancer cells in blood with over 96% accuracy. The system measured 16 biophysical parameters of cells simultaneously in single shot and performed hyper-dimensional classification using a Deep Neural Network (DNN). The results were compared with other machine learning classification algorithms such as logistic regression and naive Bayes with highest accuracy obtained with deep learning. This was later extended "Deep Cytometry" in which the computationally intensive tasks of image processing and feature extraction prior to deep learning were avoided by directly feeding the time-stretch linescans, each representing a laser pulse into a deep convolutional neural network. This direct classification of raw time-stretched data reduced the inference time by orders of magnitude to 700 micro-second on a GPU accelerated processor. At a flow speed of 1 m/s the cells only move less than a millimeter. Therefore, this ultrashort inference time is fast enough to for cell sorting.
Background
Fast real-time optical imaging technology is indispensable for studying dynamical events such as shockwaves, laser fusion, chemical dynamics in living cells, neural activity, laser surgery, microfluidics, and MEMS. The usual techniques of conventional CCD and CMOS cameras are inadequate for capturing fast dynamical processes with high sensitivity and speed; there are technological limitations—it takes time to read out the data from the sensor array and there's a fundamental trade-off between sensitivity and speed: at high frame rates, fewer photons are collected during each frame, a problem that affects nearly all optical imaging systems.
The streak camera, used for diagnostics in laser fusion, plasma radiation, and combustion, operates in burst mode only (providing just several frames) and requires synchronization of the camera with the event to be captured. It is therefore unable to capture random or transient events in biological systems. Stroboscopes have a complementary role: they can capture the dynamics of fast events—but only if the event is repetitive, such as rotations, vibrations, and oscillations. They are unable to capture non-repetitive random events that occur only once or do not occur at regular intervals.