Fig. 1. Genomics-based precision oncology to functional precision oncology^[71]

Fig. 2. Primary tumor models for phenotype function testing^[68-83]

Fig. 3. Advantages of tumor precision medicine based on optical microscopy imaging phenotype detection

Fig. 4. Multiplex immunofluorescence (mIF) imaging provides in-depth exploration of the tumor microenvironment and cell-cell interactions. (a) Multiplex immunofluorescence meets the needs of tumor spatial biology research at different research stages^[86]; (b) biomarkers can be identified in the tumor microenvironment through multiplex immunofluorescence imaging to predict responses to immunotherapy^[86]; (c) assessment of PD-L1 co-expression in breast cancer tissue microarrays (TMA) by multiplex immunofluorescence is significant for predicting responses to immune checkpoint inhibitor therapy^[87], with scale bar of 200 μm; (d) 9-color composite images presented by multiplex immunofluorescence provide a powerful tool for studying cell interactions and the tumor microenvironment^[92], with scale bar of 100 μm

Fig. 5. Autofluorescence microscopy reveals drug intervention responses in tumor cells. (a) Detecting drug sensitivity at the cellular and organoid levels based on autofluorescence from the reduced form of NAD(P)H^[94], with scale bar of 400 μm; (b) two-photon fluorescence microscopy for detecting collagen, lipofuscin, and flavin autofluorescence in tissue models can predict tumor responses to chemotherapy and immunotherapy^[95], with scale bar of 24 μm; (c) obtaining lipofuscin autofluorescence lifetime parameter through two-photon fluorescence imaging and fluorescence lifetime imaging (FLIM) techniques can distinguish between apoptosis and necrosis in individual cell level^[96], with scale bar of 50 μm

Fig. 6. Fluorescence in situ hybridization (FISH) imaging detects biomarkers for targeted therapy and related tumor molecular characteristics. (a) Amplification patterns of the oncogene MET can be distinguished using fluorescence in situ hybridization technology to analyze the relationship between this gene amplification pattern and tumor biological behavior^[102]; (b) fluorescence in situ hybridization microscopy can detect and analyze genetic abnormalities such as rearrangements, fusions, amplifications, and deletions in various tumor tissues including lung cancer, gliomas, and breast cancer^[103]

Fig. 7. Second harmonic generation (SHG) microscopy reveals information about extracellular matrix components such as collagen to elucidate tumor staging and metastatic potential. (a) SHG microscopy images illustrate that, compared to proliferating tumor cells, dormant tumor cells have a lower degree of linear alignment and higher directionalality of extracellular matrix collagen fibers^[106], scale bar: 100 μm; (b) SHG-based microscopy image reveals that in situ tumors with a propensity for metastasis have a significantly higher area of type I collagen (Col1) than subcutaneous tumors with less metastatic potential, and this distinction is more pronounced in the normoxic regions of the tumors^[107], scale bar: 60 μm; (c) SHG images of normal tissue, adjacent normal tissue, malignant tissue without metastasis, and malignant tissue with metastasis show that as the malignancy of the tumor increases, the aspect ratio of the fibers increases significantly^[107], scale bar: 60 μm

Fig. 8. Stimulated Raman scattering (SRS) imaging provides deeper insights for precision medicine in cancer. (a) SRS-based live cell imaging elucidates the relationship between cell phenotype and genotype, explaining the metabolic reprogramming mechanisms of melanoma cells, revealing potential targets for metabolic intervention therapy^[115], scale bar: 20 μm; (b) SRS observes metabolic shifts from glucose to fatty acid dependence in cisplatin-resistant cells, and metabolic index derived from quantitative analysis of SRS images can be used for tumor resistance prediction^[116], scale bar: 20 μm; (c) the workflow of Raman2RNA (R2R) encompasses cell culturing, molecular vibrational energy level detection of cells, single-molecule RNA fluorescence in situ hybridization (smFISH) imaging, parallel single-cell RNA sequencing (scRNA-seq) of cultured cells, and the prediction of single-cell RNA-seq expression profiles from Raman spectra through machine learning and multi-modal data integration^[120]

Fig. 9. Mid-infrared (MIR) photothermal microscopy and transient absorption (TA) microscopy techniques provide subcellular biochemical molecular imaging. (a) A three-dimensional view of mid-infrared photothermal imaging of PC-3 prostate cancer cells, clearly showing individual lipid droplets within the cells^[135]; (b) MIP imaging results of MIA PaCa-2 pancreatic cancer cells treated with JZL184, with images quantitatively displaying the distribution of intracellular drug and lipid content derived from multivariate curve resolution (MCR) analysis^[135], scale bar: 20 μm; (c) multicolor mid-infrared photothermal imaging of SJSA-1 cells pretreated with Dox (doxorubicin) and a cyanine-labeled enzyme activity probe, visualizing the distribution of phosphatase and caspase-3/7 activity (from left to right are brightfield, protein, lipid droplets, phosphatase, caspase-3/7, and the merged enzyme activity images)^[136], scale bar: 40 μm; (d) transient absorption microscopy imaging the AuNPs (gold nanoparticles) probe bound to Her2 mRNA, demonstrating the expression levels and localization patterns of Her2 mRNA in MCF-7 and SK-BR-3 breast cancer cells^[137], scale bar: 20 μm