Fig. 1. Scheme and principle of SRS. (a) Energy diagram of SRS and spontaneous Raman; (b) SRL detection scheme; (c) system design of SRS microscope

Fig. 2. Stimulated Ramam loss spectrum, spontaneous Raman spectrum and coherent anti-Stokes Raman spectrum^[28]

Fig. 3. Hyperspectral SRS imaging^[30]. (a) Schematic of SRS spectroscopy setup;(b) spontaneous Raman spectra of oleic acid, acetone and ethanol; (c) calibration of Raman shift with respect to pump-Stokes pulse delay using a set of Raman peaks of acetone and ethanol, inset showing the schematic illustrating the relationship between Raman shift and interpulse delay; (d) SRS spectroscopy measured by swept-delay chirped femtosecond pulses

Fig. 4. SRS images of Hela cells at different Raman shift and SRS spectra of local region. (a-c): SRS images of cells at different Raman shift; (d) overlay of protein and lipid; (e) SRS spectra of lipid and core in Fig.(b) and Fig.(c); (f) spontaneous Raman spectra of oleic acid and bovine serum albumin

Fig. 5. Hyperspectral stimulated Raman imaging system with fast scanning optical delay line^[32].(a) System optical path; (b) overhead diagram of detailed RSODL structure; (c) three-dimensional diagram of RSODL

Fig. 6. Two-color SRS images of human GBM xenograft mouse brains^[34]. (a) SRS and H&E images of a normal brain frozen section; (b) SRS and H&E images of a GBM infiltrated brain frozen section; (c) in the field of view where bright field microscope appears grossly normal, however SRS image shows a distinct margin between tumor and normal brains

Fig. 7. SRS image of normal human brain surgical tissues^[35]

Fig. 8. SRS spectra of the plaque and normal tissue^[36]

Fig. 9. Multicolor SRS images acquired on frozen AD mouse brain section^[36]. (a)-(c) SRS images at different wavenumbers; (d) three channel pseudo color composite showing the distribution of lipid (green), normal protein (blue), and amyloid plaque (cyan)

Fig. 10. Multicolor SRS image acquired on fresh AD mouse brain (lipid: green; normal protein: blue; amyloid plaque: magenta)^[36]

Fig. 11. SRS imaging of unprocessed fresh larynx surgical tissues^[37](lipid:green; protein:blue). (a)-(c) Normal squamous cells imaged at various locations of the epithelium layer;(d) enlarged cell nucleus and abnormal nuclear morphology

Fig. 12. Schematics of dual-phase SRS^[38]. (a) Setup of dual-phase SRS signal generation; (b) principle of dual-phase SRS signal generation

Fig. 13. In vivo two-color SRS microscopy images of live animals^[38](red: red blood cell; cyan:protein).(a) Two-color SRS image of heart of a zebrafish embryo; (b)two-color SRS image of blood streams in a mouse ear

Fig. 14. Illustration and demo of line scan and strip mosaicing^[39]. (a) Illustration of line scan and strip mosaicing; (b) strip mosaicing SRS image of mouse brain frozen section(lipid: green; protein: blue)

Fig. 15. Construction and validation of deep-learning model^[37]. (a) Network architecture of ResNet34; (b) schematic illustration of work flow for training and validation of the model; (c) five-fold cross validation results

Fig. 16. SRS histology of larynx tissue with the aid of ResNet34^[37](lipid: green; protein: blue; neoplastic: red; gray: normal). (a) Imaging and prediction results of a laryngeal squamous-cell carcinoma tissue; (b) results of a normal laryngeal tissue

Fig. 17. Diagnostic results of untrained cases using ResNet34^[37]. (a) Diagnostic results of 33 independent cases using ResNet34 vs. true pathology results;(b) ROC analysis of the results from ResNet34 (AUC: area under the curve)