Fig. 1. Principle of the Fourier-domain-compressed optical time-stretch quantitative phase imaging flow cytometry. The upper part shows the time-domain waveforms of each process, while the lower part displays the frequency-domain representation of the compressed signal processing. Time-stretch: the short pulse is stretched in the time domain to map the spectrum into a temporal data stream. Interference: the time-stretched reference pulse beats with the time-stretched signal pulse to form an interfered signal. Compressed sampling: the interfered pulse is modulated by the sinusoidal wave with different frequencies and then is compressed in the time domain to collect the high-frequency Fourier coefficients.

Fig. 2. (a) Workflow of cell detection with the Fourier-domain-compressed optical time-stretch quantitative phase imaging flow cytometry. (b) Detailed flowchart for sample preparation. (c) Experimental setup of the Fourier-domain-compressed optical time-stretch quantitative phase imaging flow cytometry. SMF: single-mode fiber; AWG: arbitrary waveform generator; MZM: Mach–Zehnder modulator; OL: objective lens; DCF: dispersive compensation fiber; PD: photodetector; and OSC: oscilloscope. (d) Architecture of the Inception-V3 neural network.

Fig. 3. Static performance of the imaging system. (a) and (b) Intensity and phase images of the polystyrene microsphere and corn root cross section under different compression ratios, respectively. (c) and (f) Phase curves along the dashed lines in (a) and (b), respectively. (d) and (e) Average value and standard deviation of SSIM and PSNR for the polystyrene microsphere intensity and phase images under different compression ratios. (g) and (h) SSIM and PSNR for the corn root cross section intensity and phase images under different compression ratios. Exp: experiment. Scale bar: 10 μm.

Fig. 4. Imaging performance of flowing cells. (a) Intensity and phase images of the breast cancer cells under different compression ratios. (b)–(e) Average value and standard deviation of NIQE and SEN for the intensity and phase images acquired under different compression ratios. Scale bar: 10 μm.

Fig. 5. (a) and (b) Uncompressed and compressed MCF-7 cell images under drug concentrations of control, 50 μM, and 100 μM, respectively. (c) and (d), (e) and (f) The biophysical phenotypic single indicator analysis of uncompressed cell images and compressed cell images. (g)–(i) The dual index combination analysis of uncompressed intensity images, uncompressed intensity and phase images, and compressed intensity and phase images, respectively.

Fig. 6. Performance of MCF-7 cells classification. Each row, from left to right, represents the classification results of uncompressed intensity images, uncompressed phase images, uncompressed intensity and phase images, and compressed intensity and phase images. (a)–(d) The t-SNE diagram of drug-treated and drug-untreated MCF-7 cells. (e)–(h) ROC curves for the classification of drug-treated and drug-untreated MCF-7 cells. (i)–(l) Confusion matrix for the classification of drug-treated and drug-untreated MCF-7 cells.

Fig. 7. Biophysical phenotypic single indicator analysis of uncompressed cell images and compressed cell images. (a) and (b) The image contrast analysis and image entropy analysis of uncompressed cell intensity images. (c) and (d) The analysis of dry mass and dry mass density of uncompressed phase images. (e) and (f) The image contrast analysis and image entropy analysis of compressed cell intensity images. (g) and (h) The analysis of dry mass and dry mass density of compressed phase images.

Fig. 8. (a) The biophysical phenotypic combination analysis of image contrast index and roundness index extracted from uncompressed intensity images. (b) The combination analysis of image entropy index and image contrast index extracted from uncompressed intensity images.

Fig. 9. (a) The biophysical phenotypic combination analysis of dry mass index and image contrast index extracted from uncompressed phase and intensity images, respectively. (b) The biophysical phenotypic combination analysis of the mean of phase index and roundness index extracted from uncompressed phase and intensity images, respectively.

Fig. 10. (a) The biophysical phenotypic combination analysis of dry mass index and image contrast index extracted from compressed phase and intensity images, respectively. (b) The biophysical phenotypic combination analysis of the mean of phase index and roundness index extracted from compressed phase and intensity images, respectively.

Fig. 11. Correlation analysis of the training data of the model: (a) uncompressed cellular data and (b) compressed cellular data.