Cytometry system and method for processing one or more target cells from a plurality of label-free cells
The present disclosure provides methods and systems for ghost cytometry (GC), which may be used to produce an image of an object without using a spatially resolving detector. This may be used to perform image-free ultrafast fluorescence “imaging” cytometry, based on, for example, a single pixel detector. Spatial information obtained from the motion of cells relative to a patterned optical structure may be compressively converted into signals that arrive sequentially at a single pixel detector. Combinatorial use of the temporal waveform with the intensity distribution of the random or pseudo-random pattern may permit computational reconstruction of cell morphology. Machine learning methods may be applied directly to the compressed waveforms without image reconstruction to enable efficient image-free morphology-based cytometry. Image-free GC may achieve accurate and high throughput cell classification as well as selective sorting based on cell morphology without a specific biomarker, which have been challenging using conventional flow cytometers.
1. A method for processing one or more target cells from a plurality of label-free cells, comprising:
(a) obtaining spatial information of the plurality of label-free cells using a detector, wherein obtaining the spatial information occurs during motion of the plurality of label-free cells relative to a patterned optical structure;
(b) compressively converting the spatial information into one or more temporal wave forms, wherein the one or more temporal waveforms comprise one or more intensity distributions imparted by the patterned optical structure; and
(c) applying one or more machine learning classifiers to the one or more temporal waveforms to identify the one or more target cells from the plurality of label-free cells.
2. The method of claim 1 , further comprising, subsequent to (c), separating or isolating the one or more target cells from the plurality of label-free cells based at least in part on the one or more target cells identified in (c).
3. The method of claim 2 , wherein separating or isolating the one or more target cells from the plurality of label-free cells occurs at a rate of at least 10 cells per second.
4. The method of claim 2 , wherein separating or isolating the cells further comprises sorting the plurality of label-free cells into one or more groups of sorted cells based on morphologies of the plurality of label-free cells.
5. The method of claim 4 , wherein separating or isolating the label-free cells further comprises collecting one or more of the groups of sorted label-free cells to generate an enriched cell mixture.
6. The method of claim 5 , further comprising subjecting one or more cells of the one or more groups of sorted cells to one or more assays.
7. The method of claim 4 , wherein the sorting the plurality of label-free cells comprises classifying or sorting the plurality of label-free cells without image reconstruction.
8. The method of claim 1 , wherein (a) comprises:
(i) directing a light from a light source through the patterned optical structure,
(ii) directing the light from the patterned optical structure to the plurality of label-free cells, and
(iii) directing the light from the plurality of label-free cells to the detector.
9. The method of claim 1 , wherein (a) comprises:
(i) directing a light from a light source to the plurality of label-free cells,
(ii) directing the light from the plurality of label-free cells through the patterned optical structure, and
(iii) directing the light from the patterned optical structure to the detector.
10. The method of claim 1 , wherein the one or more machine learning classifiers are created using support vector machines, random forest, artificial neural networks, convolutional neural networks, deep learning, ultra-deep learning, gradient boosting, AdaBoosting, decision trees, linear regression, or logistic regression.
11. The method of claim 1 , wherein the one or more target cells comprise one or more cancer cells or circulating tumor cells.
12. The method of claim 1 , wherein the one or more target cells comprise one or more therapeutic cells.
13. The method of claim 1 , wherein the one or more target cells comprise one or more members selected from the group consisting of: stem cells, mesenchymal stem cells, induced pluripotent stem cells, embryonic stem cells, cells differentiated from induced pluripotent stem (iPS) cells, cells differentiated from embryonic stem cells, genetically engineered cells, blood cells, red blood cells, white blood cells, T cells, B cells, natural killer cells, chimeric antigen receptor T cells, chimeric antigen receptor natural killer cells, cancer cells, and blast cells.
14. The method of claim 13 , wherein the one or more target cells comprise one or more white blood cells selected from the group consisting of: neutrophiles, eosinophils, basophils, monocytes and lymphocytes.
15. The method of claim 1 , further comprising classifying the one or more target cells based on a function, a type, or a characteristic of the one or more target cells, wherein the function, type, or characteristic corresponds to a liveliness, an expression state, a purity, a differentiation, an undifferentiation, a glucose level, or a glycolysis level of the one or more target cells.
16. The method of claim 1 , further comprising collecting a partial transmissive speckle pattern of the plurality of label-free cells as the plurality of label-free cells move relative to the patterned optical structure.
17. The method of claim 1 , wherein the patterned optical structure comprises an ordered patterned optical structure, a disordered patterned optical structure, an aperiodic patterned optical structure, a random or pseudo-random patterned optical structure, or a static optical structure.
18. A system for processing one or more target cells from a plurality of label-free cells, comprising one or more processors configured to:
(a) obtain spatial information of the plurality of label-free cells using a detector, wherein obtaining the spatial information occurs during motion of the plurality of label-free cells relative to a patterned optical structure, wherein the plurality of label-free cells flow with fluids in a fluid path;
(b) compressively convert the spatial information into one or more temporal wave forms, wherein the one or more temporal waveforms comprise one or more intensity distributions imparted by the patterned optical structure; and
(c) applying one or more machine learning classifiers to the one or more temporal waveforms to identify the one or more target cells from the plurality of label-free cells.